Sojourner: An Insider's View of the Mars Pathfinder Mission 0425191990, 2003061754

Citation preview

SOJOURNER c^P :**-(

f

/An Insiders View of the

FMars

Pathfinder Mission

¥

#

*1m

Andrew Mishkin Senior Systems Ensi

USA $21.95 Canad

^* From Andrew Mishkin. a senior systems engineer at the Jet Propulsion Laboratory

and a leader of NASA's robotic program,

comes an

look at the most and audacious space project to date—the Mars Pathfinder probe that insider's

fascinating

electrified the world's imagination.

Far out of reach of her

$

human creators— one

hundred twenty-two million miles from

home—

a sophisticated, thinking robot smaller than a

microwave oven crept among the rocks of an at

alien landscape. Six-wheeled Sojourner was doing

what had never been done before— exploring the rough, red terrain of Mars. Soon, she would

beam

spectacular pictures of her one-of-a-kind travels

back to

Earth.

And

millions of people would

be

captivated.

Now, with the touch of an expert thriller writer, Sojourner operations team leader Andrew

Mishkin

tells

the inside,

human story of the Mars

Pathfinder team's feverish efforts to build a

self-

guided, off-roading robot to explore the surface of

the Red Planet. With witty, compelling anecdotes,

he describes the clash of temperamental geniuses, the invention of a turf wars, the

new work ethic, the

chewing-gum solutions

to high-tech

problems, the controlled chaos behind the strangely beautiful creation of an

artificial

intelligence— and the exhilaration of inaugurating

the next great age of space exploration.

SOJOURNER

Most Berkley Books

are available at special quantity discounts for bulk purchases for sales

promotions, premiums, fund-raising, or educational use. Special books, or book excerpts, can also be created to

For

fit

specific needs.

details, write: Special

New York, New York

Markets,

10014.

The Berkley Publishing Group, 375 Hudson

Street,

SOJOURNER AN INSIDER'S VIEW OF THE

MARS PATHFINDER MISSION

Andrew Mishkin

i: BERKLEY BOOKS,

NEW YORK

A Berkley Book Published by The Berkley Publishing

Group

A division of Penguin Group (USA) 375

Hudson

New York, New York, This

book is an

original publication of

Copyright

Inc.

Street

10014

The Berkley Publishing Group.

© 2003 by Andrew Mishkin.

Text design by Tiffany Estreicher.

All rights reserved.

This book, or parts thereof,

The

may not be reproduced in any form without permission.

scanning, uploading, and distribution of this

without the permission of the publisher

is

illegal

book via

the Internet or via any other

means

and punishable by law Please purchase only

authorized electronic editions, and do not participate in or encourage electronic piracy of

copyrighted materials. Your support of the author's rights

BERKLEY and the

is

appreciated.

"B" design are trademarks belonging to Penguin

First edition:

Group (USA)

Inc.

December 2003

Library of Congress Cataloging-in-Publication Data

Mishkin, Andrew.

Sojourner

:

an

insider's

view of the Mars Pathfinder mission p.

/

Andrew Mishkin.

—

1st ed.

cm.

Includes index.

ISBN 0-425-19199-0 1

.

Mars Pathfinder Project 4.

(U.S.).

Mars

2.

Sojourner (Spacecraft).

(Planet)

— Exploration.

TL789.8.U6P386

I.

3.

Title.

2003

629.43'543—dc22 2003061754

PRINTED IN THE UNITED STATES OF AMERICA

10

987654321

Space

flight to

Mars.

For Hank Moore,

who

loved Sojourner most of all

CONTENTS

The Solar System and Mars

ix

Preface

xiii

Prologue

l

Part

LAYING THE

1:

GROUNDWORK

1.

Doing What's Never Been Done

2.

"Almost As Good

3.

Off-roading with No One at the Wheel

25

4.

The Right Place at the Right Time

38

5.

The Big Rover That Never Would

52

6.

The

65

Little

Asa

Bogie"

Rover That Could

?

13

Part 2:

PATHFINDER A Small Enough Team to Do the Job

85

8.

The Rover War

9?

9.

A Design That Really Works

10?

Three Rovers

124

?.

10.

Contents

viii

11.

Seeing and Believing

134

12.

Two Spacecraft

145

13.

Trial

by Centrifuge

155

14.

Can

We Talk?

165

15.

The Noise That Wouldn't Die

i?5

16.

Soul of Sojourner

185

l?.

License to Drive

195

18.

Meteorites,

Life,

and Job Security

20?

Part 3:

GOING TO MARS 19.

Even a Journey of a Thousand Miles ...

21?

20.

Cruisin'

223

21.

So What Are You Going to Do for the

Next Six Months?

232

22.

Testing, Testing ...

244

23.

Moments

261

24.

Living on Mars Time

25.

Will Build

of Truth

Spacecraft for Food

282

304

Epilogue

311

Glossary

313

Dramatis Personae

316

Acknowledgments

318

Index

321

Photographic Credits

334

THE SOLAR SYSTEM AND MARS

Beyond the Moon, Earth's closest planetary neighbors are Venus and Mars. With our current technological capability, robotic spacecraft

can reach either of these worlds in only months of

—Mercury, Venus, creasing distance from the sun— are of the inner planets

Earth, and

solid,

half a mile

up

now

Pluto.

in order of in-

asteroid belt, remnants of

consisting of thousands of pieces ranging

to hundreds of miles across. This field of debris

the transition to the outer planets

four

rocky worlds. Traveling out-

ward from Mars, you would next encounter the once larger bodies,

Mars

travel. All

from

marks

—Jupiter, Saturn, Uranus, Neptune, and

These worlds are years away

at the

speeds our spacecraft can

achieve. Except for Pluto, the outer planets are

and gaseous worlds with no discernible of the solar system have one or many

all

gas giants, huge liquid

solid surfaces.

Most of the

planets

moons orbiting them; some of these

moons rival planets in size, while others are mere chunks of rock and ice. The planets are varied and distinct. Mercury is small, airless, and cratered.

Venus

is

the greenhouse planet, with a crushingly dense carbon

dioxide atmosphere and a surface temperature hot

The

enough

largest planet in the solar system, Jupiter, could

to melt lead.

swallow thirteen

hundred Earths. Rings thousands of miles across but only 450 girdle Saturn.

Uranus

is

a gas giant planet turned

on

its side,

feet thick

with one of

The Solar System and Mars

x

its

poles sometimes pointing nearly directly at the sun. Blue-banded Nep-

tune

is

nearly 3 billion miles from the sun, thirty-one thousand miles in

ameter, with storms in

its

atmosphere

of the

known

is

Pluto, solid

Moon,

so far

planets

away

that,

as big as planet Earth.

together with

around the sun only once in 248 Earth

and its

icy

Most remote

and smaller than Earth's

own moon

Charon, Pluto goes

years.

Far beyond the planets (months away even at the speed of the deep darkness

where they cannot be

sun: perhaps billions of icy comets,

ices into a visible tail millions

light), in

detected, other objects orbit the

composed of material unchanged

since the formation of the solar system.

into the realm of the inner planets,

di-

Only

rarely does a

comet venture

where the heat of the sun vaporizes

of miles in extent.

Neptune

The

orbits

and

relative positions

of the planets on July

4, 1997.

its

The Solar System and Mars

xi

Our Earth orbits the sun at a range of about 93 million miles.

among the

It is

unique

planets in having an atmosphere of mostly nitrogen and oxy-

gen, and oceans of water covering 70 percent of

surface. Earth's

its

Moon

has no atmosphere, always shows the same face toward Earth, and

bombardment from

cratered with the evidence of meteor

is

of

billions

years ago.

Half again

as far

This reddish world

with

from the sun as Earth

misshapen moons diminutive

size,

—Mars.

smaller than Earth, with only a tenth the mass, and

is

entire surface

its

orbits the fourth planet

about the same area

—Phobos and Deimos—

Mars

is

the

as Earth's continents.

circle the planet.

Two

Despite

its

home of Olympus Mons, the largest identified

volcano in the solar system

—

—and of

rising fifteen miles into the sky

Valles Marineris, the grandest canyon, as long as the continental United States

wide.

is

Much

of the planet

is

covered with impact craters and vol-

canoes. There are also channels that appear to have been created by flow-

ing water and catastrophic floods. Yet there the surface today.

The this

Where

of Mars

air

atmosphere

is

pressure of Earth's

is

is

no

liquid

water

visible

on

did the water go?

composed almost

entirely of

carbon dioxide. But

very thin, only about one two-hundredth the sea-level air.

Although the atmosphere

is

tenuous, winds of up

to 180 miles per hour sometimes result in regional and even planet- wide

dust storms that obscure surface features for gravity

is

weaker than

Earth's,

like Earth,

thirty-eight

and

is

just a

pounds on Mars.

as the seasons change,

caps (of carbon dioxide and water ice)

day

grow and

touch longer than Earth's,

at

ebb.

rise to as

but even in the more temperate equatorial regions

they

is

it

polar ice

may

high as 60 °F,

fall at

-90 °F, with temperature swings of up to 110° F in a single greater distance from the sun causes

its

The mean Martian

twenty-four hours and thirty-

nine minutes. During the day, the temperature can

year that

Martian

at a time.

such that a person weighing a hundred

pounds on Earth would weigh only

Mars has seasons

months

to orbit

more

night to

day. Mars's

slowly, giving

it

a

687 Earth days long. As Earth and Mars travel in their courses,

may pass as close to each other as 34 million miles; but when their or-

bits take

them

to opposite sides of the sun, the distance can be as great as

230 million miles.

— The Solar System and Mars

xii

Some have postulated that Mars was once much more due to

its

smaller size, aged

planet of today But thrived

on the

quickly to

no one knows

become

for sure that

planet, or that vestiges of such

Mankind has turned beyond the

more

Moon—it is

the frontier of space

.

.

.

life

life

like Earth,

but

the seemingly dead

might not have once

might not yet remain.

its

attention to the exploration of Mars, for

the

most hospitable and reachable destination

in

PREFACE

he story of Sojourner

T

was

small, but

for a

number of years

ing, there

with the I

is

no

it

is

a

broad and complex one. The rover

to

single reality shared

by

all

involved.

The

story changes

storyteller.

have reconstructed the history of Sojourner with the aid of notes,

conversations, interviews, emails, official documents, and personal ories. In

own

itself

many people who invested their lives make it work. As with any grand undertak-

represented

some

cases

I

have attempted to re-create conversations from

recollections or those of others. For brevity

sented such conversations in quotes.

I

and

have always had

style,

the content of the situation accurately,

The

when

exact

I

at least

to the original conversation review the text to establish that

available.

mem-

words

my

have pre-

one

part}'

it

represents

are

no longer

responsibility for any inaccuracies or misrepresentations

introduced must of course be

my

own.

Some of them Some devoted full time and more for all of the years of the Pathfinder mission. Some worked part time, or contributed a necessary element and moved on to the next job that needed

Many

people contributed to the success of Sojourner.

appear in the chapters that follow.

-

doing. sight:

The nature of

The

"faster, better,

cheaper'

is

that there

is

less over-

success of the entire mission depends on each individual doing

xiv

Preface

his or

her job well. Despite

my

involvement with the development and

operation of Sojourner, any story

tell

I

must by necessity represent only

limited interpretation of the facts as colored

by

my personal

Most members of the rover team do not appear count, and tions.

I

have

little

However, the

personal knowledge of

fact

in the

many

of their omission from

a

experience.

body of

this ac-

of their contribu-

this text

should not be

construed to diminish their importance to the mission.

A history of Sojourner must be the story of two journeys:

the rover's

voyage to Mars, and the shared experience of the individuals who,

made

together,

that voyage a reality In an attempt to ensure that the

members of the rover team receive the recognition they are due, I have included below the most complete compiled list of members of which I am aware.

I

have not tried to encompass the entire Pathfinder mission team,

but only the individuals, institutions, and companies directly participating in the rover effort.

apologies.

If,

despite

my

best efforts, errors remain,

To the rover and Pathfinder teams

I

offer

my

privilege of being a part of a once-in-a-lifetime adventure.

I

offer

my

thanks for the

We did

it!

THE ROVER TEAM

Jet Propulsion Laboratory

Robert D. Galletly

JackC. Morrison

George A. Alahuzos

Jeffrey A. Gensler

Robert

L.

Ghanim Al-Jumaily

James A. Gittens

Fred

Nabor

Teresa D. Alonso

Willis

Sami W. Asmar

Christopher Hartsough

Lawrence

W

Ronald

Banes

S.

Edmund

Avril

C. Baroth

W Han

Gregory

Mueller

Michael A. Newell

Tarn

T

Nguyen

Don E. Noon

Hickey

S.

F.

Ohm

Denise A. Hollert

Timothy

Sean Howard

Argelio Olivera

Hua

Thomas

R.

Sheryl L. Bergstrom

Tien

Donald

Kenneth A. Jewett

Mervin

Russell D. Billing

Kenneth

Michael D. Parks

David J. Boatman

Vaughn J. Justice

Douglas C. Perry

Stephen R. Bolin

Faramarz N. Keyvanfar

Mark Phillips

Gary

Eug-Yun Kwack

Betty L. Preece

B. Bickler

Bolotin

S.

Thomas J. Borden David

Braun

F.

Kyle D.

Brown

Robert

E.

Brown

Sharon

R.

L.

Johnson

Thomas A. Rebold

Laubach

Evelyn Reed

Geoffrey A. Laugen E.

Todd

Litwin

E.

Andrew

Layman

William

O'Toole, Jr.

K. Parker

Robert

D. Rose

E. Scott

Dale R. Burger

James W. Lloyd

Terri A. Scribner

John M. Cardone

Gena Lofton

Leonard A. Sebring

Brian K. Cooper

Justin N.

Cosme M. Chavez

Ramachandra Manvi

Joy A. Crisp

William H. Mateer

Evan D. Davie

Jacob R. Matijevic

Allen R. Sirota

Larry H. Matthies

Hugh Smith

Henry Delgado

Thomas M. McCarthy

Kathy Sovereign

Fotios Deligiannis

Robert McMillan

Beverly

Tolis Deslis

Donald

William C. Dias

Robert J. Menke

Bob

C.

Debusk

Dee

Maki

P.

II

McQuarie

Cesar A. Sepulveda

Donna

Marian G. Meridiem

Philip

Andrew H. Mishkin

Henry

Howard J.

David

Eisen

S.

Gordon

Mittman R.

Mon

St.

Ange B. Stell

Paul M. Stella

Patrick L. Dillon

Ellers

L. Shirley

Christopher

Johnny Duong Khanara

Sedgwick

R.

P.

Stevens

W Stone

Scot L. Stride

Lee

F.

Sword

Robert G. Moncada

Hung Ta

Harvey A. Frank

Henry J. Moore

Jan A. Tarsala

Jack A. Frazier

Yvonne Morales

Brenda

Fieri

Bert H. Fujiwara

Ronald A.

Morgan

Robert

F.

Thomas

Arthur D. Thompson

Joseph

Toczylowski

F.

Pasquale

DiDonna

Lucas Schaevitz

Economou Kubo

Peter Tsou

Thanasis E.

Maxon

Lin M. van Nieuwstadt

Holly A.

Minco

Matthew

Maury

T.

Wallace

Wilson "Bud" Richard

F.

Perkins

Michael J.

Yolanda Walters

Watkins

Phillips

Liang-Chi

Wen

Brian H. Wilcox

Motorola Derivative Technologies Division

Rudi Rieder

Motorola Radius

Contractors

National Semiconductor

V Welch

George H.Wells, Jr.

Precision Motors

Division

Astro Aerospace

Corporation

Pacific Scientific

Pico Electronics

Larry Wild

Beckman Instruments

Pioneer Circuits,

Paul B. Willis

Canon Connectors

Power Trends

Rosalinda Wilson

Castrol

Rosemount Aerospace

Data Radio

Inc.

Lewis Research Center

Department of Energy

Dale C. Ferguson

Environmental Test

Phillip

P.

Jenkins

Joseph C. Kolecki Geoffrey A. Landis

Robert J. Makovec

Laboratory

Eastman Kodak/ Microelectronics

Technology Division

Lawrence G. Oberle

Falcon Designs

Steven M. Stevenson

Garwood

David Wilt

Globe Motors

Laboratories

Hewlett-Packard Van

APXS Instrument Team Richard

S.

Blomquist

Frank DiDonna

Inc.

Nuys Interelectric A.G.

Litronic Industries

Saft

America

Spectra Diode

Laboratory Systron

Donner

Tecstar/ Applied Solar

Division

Telogy

Wyle Laboratories

PROLOGUE

1985

April

"I

can't claim to

The the

two

about

be an expert on

light in the office

interviewers.

it

The

Artificial Intelligence."

was dim.

I

was

dark-haired

splitting

my

attention

man to my left smiled.

between

"We'll learn

together," Neville Marzwell said with an indeterminate accent.

He seemed kind, Ed Kan,

yet

somehow

mysterious.

sitting directly across the table

from me, was ready

to finish

He had a project that should have started some time before, but Marzwell had people working his own project, but needed more.

things up. hadn't.

"We'd It I

like

you

to start

Monday/'

my second interview at the Jet Propulsion Laboratory. at my own company for about three years. My were mostly brilliant but temperamental engineers. We had

was

had been working

partners

been designing hardware, doing software consulting, and writing venture capital proposals that never got funded. lar

paycheck,

I

was

One

I

decided

I

wanted

a regu-

began sending out resumes and wondering who would

hire a systems engineer a

since

When

few years out of school whose only experience

as a struggling entrepreneur.

of

my first interviews was at JPL in Pasadena.

I

got a

call

from an

SOJOURNER

2

old friend a

worked with when we were both graduate students

I'd

at

UCLA

He was moving to the east coast, and his position in the

few years before.

JPL Advanced Teleoperation group was opening up. This group worked on Space Shuttle flight experiments, designed robotic manipulators to be operated remotely either by astronauts in space or people on the ground,

and researched

how to

control these devices effectively from hundreds or

thousands of miles away.

and would

like to

I

My friend was recommending me for his position,

come

in

and meet with

his supervisor?

I

went

in,

met

Ed Kan and others, had lunch in the JPL cafeteria with his robotics group. The interview had seemed to go well. But JPL didn't call. I sent out more resumes. After a few weeks I made the "follow-up" call. "We decided not to fill the position," Kan told me.

Two months

later,

copies of

my resumes were

completing their me-

anderings through the labyrinths of several companies' personnel

The number of like to

do

second interview."

a

Ten days most I'd

callbacks increased.

later

I

this,

Ed Kan

called

me. "We'd

How soon? "Tomorrow."

started work.

certainly could have

Amid

offices.

I

could have waited for other

wrangled

a higher salary

always wanted to be in the Space business.

And

offers.

somewhere for that,

else.

I al-

But

JPL was the

place to be.

July

5,

1997

For the

first

time, the Sojourner rover's six cleated wheels

tracks in the butterscotch-hued Martian soil of Ares Vallis.

Pathfinder lander, Sojourner's

home

had made

Nearby

sat the

and protector during the seven-

month voyage to Mars. Inches from the rover's wheels was the end of the ramp down which it had just driven. It was mid-afternoon: a dimmer, smaller sun was halfway down the sky.

On Earth, the rover team was ecstatic. rived

from

applause.

far away.

". .

.

six

The Mission Control

wheels on

Sojourner was

soil!"

fulfilling its

Images of the rover had just area

ar-

was wild with cheers and

announced the Rover Coordinator.

promise.

We now

had

a fully functional,

Prologue

3

healthy rover on the surface of an alien world. to drive

the machine we had built and trained

that team,

we were It

I

The

rover

to operate.

team was ready As

a

member of

was both awed and humbled by what we had done, and what

about to do.

was time

to

go exploring.

PART

1

LAYING THE GROUNDWORK

ONE

A

DOING WHAT'S NEVER BEEN DONE

Out

beyond the orbit of Neptune

bright

star.

Even out

it is

cold.

this far, there are

Very

cold.

The sun

is

a

charged particles of matter

streaming from the sun. This tenuous solar wind becomes ever

weaker the farther one and

is

travels into

deep space. Eventually, the flow meets

halted by particles from outside the solar system

wind. Astronomers

As distant

call

realm

as this

made by human

the region is

this

—the

1

and

interstellar

happens the "heliopause."

from human experience, through

hands. Voyagers

from home, searching

where

it

fly artifacts

2 are traveling farther

and farther

for the heliopause, each

reports of their latest instrument observations.

week sending back

Home

is

radio

the Jet Propul-

sion Laboratory

One space."

of the key objectives of JPL

Earth, and

you

ifornia,

JPL

simply "the robotic exploration of

NASA centers launch human beings into space to orbit the have sent men to the surface of the Moon. JPL sends scouts

Other

into the solar system If

is

drive a

you

facility

will is

where humans cannot yet

go.

few miles west on the 210 Freeway out of Pasadena, Cal-

soon notice

a

complex of structures

off to the right.

The

a sprawl of buildings located at the base of the foothills of

the San Gabriel Mountains, surrounded only by residential neighbor-

hoods and undeveloped

hillsides. It is

here that missions to the planets are

SOJOURNER

8

designed, implemented, managed, and controlled. And, perhaps

renowned, here

is

where engineers troubleshoot spacecraft

most

that are

al-

ready en route to unimaginably distant destinations, and invent work-

arounds to keep missions

repairman

when components

alive

fail

and the nearest

hundreds of millions of miles away.

is

The atmosphere

at

JPL appears

The

relaxed.

campuslike,

facility is

with over a hundred buildings spread out over a large area, and a mall by

main gate with greenery and

the

The

highest, at nine stories,

the

is

There are few

fountains.

tall

buildings.

main administration building on the

north end of the mall.

JPL

is

about Space. There are a few

named for spacecraft: Road. The embroidered

that

are

forms boast Saturn and

A

displays

typical

its

rings.

On

hall

pure

floor of the Space

first

exhibit pictures of

detail in

each one representing a

its

new

maps of

its

art gallery

products, but its

of

JPL

spacecraft, the

scientific discovery.

the surface of cloud-

Venus, or volcanoes erupting into space

Despite space,

of the

been transformed into an

company might

these walls a visitor can see radar

enshrouded

In

The

images of planets and moons photographed by

unprecedented

and they

here: Surveyor Road, Explorer

shoulder patches on the security guards' uni-

Flight Operations Center has sorts.

streets that link the site,

came from

on Jupiter's moon Io.

current reputation, JPL was not established to explore

and the notion of doing so was,

at the

time of

its

birth, considered

fantasy.

1926, the California Institute of Technology received a $300,000 grant

to establish an aeronautics laboratory

and graduate school. Over the next

two years the Guggenheim Aeronautical Laboratory of the stitute

of Technology

—

known

Theodore von Karman became

von Karman and GALCIT

as

its

that JPL

One of von Karman s graduate

GALCIT

director.

would

It

—was

built,

California In-

and

was out of the

1930

activities

of

arise.

students, Frank J. Malina, proposed to

examine the problems of rocket propulsion for his dissertation. eral

in

The gen-

academic attitude of the time was that rocketry had no merit and

belonged to the realm of pseudoscience, yet von Karman okayed the

re-

Doing What's Never Been Done

search. Together with

two

9

who "wanted to

enthusiasts

fire

rockets,"

Ma-

lina assembled a rocket motor within a year. They were ready for their

on Halloween

first test

stand and the nozzle pointed up at the tion in the dry riverbed of the

the Caltech campus. tation

The

The

sky.

riverbed

was

The

first test

had

failed.

miles northwest of

six

largely devoid of

set off a rocket engine.

dled behind piled sandbags and ignited the

a test

researchers picked a loca-

Arroyo Seco, some

and seemed a good spot to

instantly.

mounted on

1936, with the rocket engine

motor

flammable vege-

The group hud-

—which shut down

So did the next several attempts. But by

January, they succeeded in operating their rocket engine for nearly fortyfive seconds.

A few GALCIT test

became

other Caltech graduate students

out of the pockets of

As World War

II

(Fifty years later,

a

off the Caltech

robotic rovers

efforts

They were

JATOs helped

first

tested in the failures,

same

of

and

spot.)

inspira-

practical rocket engines in the

called JATOs, for Jet Assisted Take-Off.

airplanes get

site

on land leased from Pasadena.

combination of theory, experimentation,

safely into the

almost entirely

campus, relocating to the

would be

growing team developed the

States.

the war,

its

members.

its

the 1936 rocket tests in the Arroyo Seco,

United

small

loomed, government funding flowed into GALCIT.

The rocket group moved

tion, the

The

rocket group continued theoretical analysis, construction and

of simple rocket engines. The group funded

Through

interested.

from the decks of

During

aircraft carriers

air.

* On October

4, 1957,

the

Union of Soviet

named

bit the Earth's first artificial satellite,

the

Army

the

first

trol

Ballistic Missile

Agency

Socialist Republics placed into or-

to

JPL partnered with respond within three months with

successful launch of a U.S. satellite

of JPL transferred from the U.S.

Sputnik.

—Explorer

Army to

1.

Soon

after,

con-

the newly created National

Aeronautics and Space Administration. JPL managers expected to continue development of

new rockets and

space propulsion systems.

But by 1960, JPL's focus had shifted almost solely to the spacecraft that

would ride those rockets

to the

Moon and the planets.

In the early years of

SOJOURNER

10

Moon and through

on the

the decade, JPL sent space probes to hard-land interplanetary space to fly past Venus.

Other JPL missions followed built for

JPL by Hughes

Aircraft

face in advance of the Apollo

in the late sixties

Company, soft-landed on the lunar

coming

Viking was the orbiters

first

was the

planet.

It

was

also the first planetary ex-

with another

NASA facility,

JPL found

pany

Mars

orbiters,

itself in a rare

the Langley Research Center in

Virginia, having ultimate responsibility for Viking

to JPL for the

spacecraft to closely

mission to reach the surface of Mars, sending two

and two landers to the

role,

first

closer to the sun than any previous probe.

ploration mission with a billion-dollar price tag.

support

sur-

human landings. The Mariner missions flew

past Venus and Mars. In 1973, Mariner 10

explore Mercury,

and beyond. Surveyor,

s

success. Langley

came

and to the Martin Marietta Aerospace Com-

for the landers.

JPL proposed the Grand Tour of the outer planets. Due to a fortuand Neptune that would take

itous alignment of Jupiter, Saturn, Uranus,

place in the late 1970s, but not again for hundreds of years, possible to send a single spacecraft skipping collecting data

from them

effect possible at

all.

it

from one planet to the

By taking advantage of the gravity slingshot

much sooner than by a

dedicated direct trajectory.

budget for the Grand Tour was not forthcoming. JPL scaled

spacecraft

would

next,

each planetary encounter, the spacecraft could reach the

next planet in line

Voyager, and did

would be

it

down

to

anyway. The Voyagers would prove so reliable that the

still

be functioning twenty-five years

Deep space missions grew billion-dollar-class

it

The

mission

ever

more

later.

ambitious. Galileo— another

—was not to just pass near

into orbit, exploring the entire Jovian system of

Jupiter,

moons. As

but to go

if this

weren't

enough, Galileo would release a probe that would parachute into Jupiter's atmosphere, dropping deeper and deeper until creasing pressure and finally vaporized

The other JPL space probes built approached Galileo bit

Venus, mapping

by the

it

was crushed by the

planet's intense heat.

in the late eighties

in complexity, cost, or ambition. its

in-

and early

nineties

Magellan was to

or-

surface. Since cameras could not see through the

dense Venusian clouds, Magellan would use radar imagery to to a resolution even better than yet achieved for Earth

by

map Venus

satellites orbit-

Doing What's Never Been Done

ing our

home

cases

would

Mars Observer was planned

planet.

planet, carrying

1

many

to circle the fourth

science instruments, taking pictures that in

distinguish features less than six feet across.

And

some

the Cassini

mission would do for the Saturnian system what Galileo would do for the Jovian.

Even

as Cassini

ment, the States

was being approved by Congress

budget was tightening.

just too

to begin develop-

and economic climate was changing. The United

political

A billion dollars for a planetary mission was

much. Weren't there

better,

more important ways

to spend the

money? During the public

ment

first

twenty-five years of the space program, the American

had perceived agency.

NASA

On January

as

perhaps the only truly competent govern-

28, 1986, this

view

faltered.

The Space

Challenger exploded seventy-three seconds after launch, killing bers of

NASAs

its

crew.

The

shuttle fleet

was grounded

for over

two

Shuttle

all

mem-

years.

But

reputation was grounded for years to come. NASA's vaunted

"safety first" policy

was made

a

lie.

Other space

failures followed, includ-

unmanned satellite launches that either blew up or failed to proper orbits. Some of these had no connection with NASA, but

ing a string of achieve

seemed to go unnoticed in the public

this distinction

Shuttles finally

ployed into

began

orbit,

flying again, the

eye.

When the Space

Hubble Space Telescope was

de-

with high hopes that it would enable astronomers to see

farther into the depths of the Universe than ever before in history. Yet only a

few weeks

after launch, the

Hubble was found

to be "nearsighted"

due

to a manufacturing error.

JPL was not immune

to high-profile problems.

The

Galileo spacecraft,

intended for launch from the Space Shuttle in 1986, was delayed for years in the aftermath

of Challenger.

When

it

was

finally

"high gain" antenna, designed to open like an umbrella,

only partially deployed.

The antenna was

useless.

Without

fraction of the planned high-resolution images of Jupiter

would ever be

sent back to Earth.

its

main

became

stuck

launched,

it,

only a small

and

its

moons

A likely cause of the problem was iden-

tified:

lack of lubrication of the antenna

neers

who

deployment assembly. The engi-

designed the antenna had not anticipated the delays in the

launch of Galileo. In the original schedule, Galileo would have launched

SOJOURNER

12

and the antenna would have been opened within and the

tion,

initial

lubrication

would have

a year of

its

construc-

sufficed. Instead, over three

years had elapsed before launch, and the lubricant had dissipated.

Mars Observer was

to be the

first

return to the

Red Planet

in fifteen

Both the launch and the cruise to Mars proceeded without incident.

years.

But just hours before the spacecraft was to into orbit

around Mars,

it

fire its

thrusters to put itself

stopped communicating with the Earth. Ground

JPL attempted to reestablish a link for months, but there was never an answer. JPL had lost a spacecraft for the first time in twenty controllers at

years.

In the middle of this

gloom, the Magellan mission was a success, map-

ping over 95 percent of the surface of Venus, outliving

expected mis-

its

sion duration.

NASA's budget had

been shrinking for years, ever since the

late

Apollo days.

Congress's support of space exploration was at best halfhearted.

NASA forts: less.

Administrator responded with a

they must be

These seemed

"faster, better,

at first to

new mantra

cheaper."

The

for future space ef-

NASA must

do more with

be empty buzzwords. Instead, they signaled

recognition of what had to be done differently perceived differently,

NASA was going to

survive.

Now JPL would have to do more with less as well. tion could build spacecraft to

Each

would no longer be enough. The winds had

The

"faster, better,

to the future of JPL?

More was

cheaper" environment

what had never been done

organiza-

But simple technical

lence

mean

The

perform unprecedented technological

mission into the solar system proved

this

if

before.

it.

riding

on

shifted. its

feats.

excel-

What would

actions than ever.

now challenged JPL to again do

TWO -It-

"ALMOST AS GOOD AS

Don

Bickler

was looking

the costs of solar

cells.

process of manufacturing solar into wafers,

at

JPL

making

cells

adjusted for inflation,

program was

cells.

it

was down

pointed. It

and the

what made "I

a

to

finally

mil-

improve the

constructing the

dollars per watt.

was

the pro-

to just a dollar a watt.

Ten years

And

for Bickler at JPL.

the

it.

In

later,

money

his interest.

About

But he kept digging around

this time, his kid

came

work, and

could do better than

how

well

it

bought

a Jeep.

across. Bickler stud-

available after-market modifications to it

to bring

hundred

Bv 1985 JPL had done

Just like he did with any mechanical system he

to see

—over

starting to dry up.

something to pique

ied the Jeep

work on ways

Bickler' s specialty

sliced wafers.

was about twenty

There was other work for

of interest in

including refining the silicon, slicing

from the wafers, and

from the

cells

1975. the cost per panel

for the

a lot

JPL sought

project.

cells,

panels from quantities of individual

duction of the

in 1975 to

The program was big

As part of the

lion dollars per year.

it

had been

'73 there

energy sources. That interest had translated into

monev. Bickler had been hired

down

something new.

embargo of

.liter the oil

alternative

for

BOGIE

A

performed.

its

suspension

He was

disap-

this."

might be intriguing to attack the problem of designing high-mobility

SOJOURNER

14

vehicles intended to

go

off-road.

Bidder began to think that maybe

it

was

time for the world to get interested in extraterrestrial vehicles. Extraterrestrial,

by

Bickler hicles.

definition,

knew

meant way

that there

Carl Ruoff,

off-road.

was ongoing research

who had been

at

JPL

into robotic ve-

supervisor of the Robotics group, had

managed to capture some of JPL's scarce internal research-and-development dollars

and

direct

them toward developing some of the fundamental techmake "planetary rovers" possible. Bickler had once

nologies required to

given Ruoff funding to buy a robot

sembly of solar

The

to experiment with robotic as-

Maybe Ruoff wouldn't mind a little help with rovers.

cells.

trouble

arm

was

that Bickler

was

in the

Mechanical Systems division,

and Ruoff worked in the Electronics and Control perennial bone of contention between the

two JPL

and applications for those technologies, gray areas

ment hierarchy were

a fact of

life.

The

electronics

was

a

organizations. In an

thousand people constantly developing

institution of six

gies

division. Robotics

new

technolo-

in the

manage-

and control engineers

thought robotics was their domain, because making a robot do the right thing required that software and control algorithms be combined with sensors and motors into a single integrated system.

The mechanical

people saw the robot as a complex electromechanical system composed of motors, gears, and linkages that needed to be designed elegantly to together: "Those control guys always

problem that would be in the first place

agement

in

.

.

."

trivial if

want

work

to write software to solve a

they had just designed the hardware right

And robotics was one of the sexy technologies. Man-

both divisions wanted to dominate robotics

activities at

JPL.

He had found an unresolved He just wanted the opportunity

Bickler didn't care about any of that.

engineering research topic to delve into. to

do

it.

Amid

the technological saber rattling, Bickler had a meeting with

Ruoff. Ruoff

was an even-tempered man with

dity of the foibles of large organizations.

a

keen sense of the absur-

(He would sometimes pat

half-

inch-thick copies of viewgraph presentations he had been forced to either sit

through or present himself, and then comment wryly: 'Ah, view-

graphs.

Our most important

wanted

to do,

Ruoff

product!")

said, "Well,

Don,

When

in spite

Bickler told

him what he

of your management,

Til

"Almost As Good As a Bogie"

1

cooperate with you." Ruoff told him to go see Brian Wilcox,

who was co-

ordinating the planetary rover research effort.

So Bickler walked across the Laboratory to Wilcox's

office.

The build-

ing was about as far from the main entrance of JPL as you could be and still

be

on-site.

That

also put

ducing the number of botics group did

Wilcox had recently

he was

its

it

from the administrative

offices,

thus

from management. This was where the Ro-

visits

tasks in skunkworks-like isolation.

a thick black

beard that covered most of his

face.

more interested in exploring creative ways to make

than in moving further up the

He had

management

things

Surveyor Lunar Roving Vehicle, or

had been constructed

in the early 1960s

work

chain.

rover researchers could not afford to build a

new

rover for their

experiments. Instead, they had refurbished a rover that was already able: the

re-

been promoted to supervisor of the Robotics group even though

far

The

far

SLRV

This

six- wheeled

and was designed to

fit

avail-

rover

onboard

one of the unmanned Surveyor spacecraft that were being soft-landed on

Moon in

the

proven that

advance of the Apollo manned landings. The Surveyors had

safe landings

would not disappear signers

had

also

sent to the

possible,

and that the

first

forever beneath an ocean of dust.

hoped

scout the surface.

were

to send a lunar rover, operated

Two

Moon. With

prototype SLRVs were a

few thousand

one SLRV out of mothballs,

installed

and made the vehicle operational

The Surveyor

to

but none were ever

team had pulled

batteries, cameras,

They

de-

from the Earth,

dollars, Wilcox's

new

again.

built,

lunar astronauts

and

tires,

also painted over the origi-

nal white color with a light blue. From that point on, the vehicle was known simply as the "Blue Rover." What Bickler wanted to do was to optimize the mobility performance of a planetary rover. Could he come up with a design that would be better

than the Blue Rover at driving over rocks and crossing crevasses?

could he minimize the chances that a rover would sink into sand?

Was

six

the right

number of wheels? He

"Is

there anything

I

can

did not

want

to

What he asked Brian Wilcox do without mowing your lawn?"

compete with the current research

was

How

get stuck on a rock or

activity.

Wilcox responded that Bickler was welcome to investigate rover mobility characteristics.

Wilcox had much more important things to worry

— SOJOURNER

16

about. His

team had never signed up

were deep into the

on Mars,

say,

issue of

for instance

pose the rover had a

how to

was so

—that there was no way to drive

TV camera mounted on

it,

it

away

far

directly.

Sup-

sending back pictures of

the terrain in front of the rover. Traveling at the speed of

on Earth would take another twenty minutes

light,

those

TV

Commands from

images could take up to twenty minutes to reach Earth. a driver

They

to improve the Blue Rover.

control a rover that

back to the

to get

rover where they were needed. So any instructions to the rover, such as

"Stop before you go over that

Time with

delay,

even

at

would

cliff!"

Somehow, the rover had to be made

it?

minutes too

arrive forty

the speed of light, could be a to

killer.

late.

How did you deal

go to the

right place, with-

human being immediately available to tell it what to do. No matter how mobile the vehicle might be, there could still be

out a

big

enough to get stuck on, and crevasses wide enough to

other hand, the rain

it

more capable

could handle

safely,

the vehicle's mobility, the

fall into.

more

rocks

On the

types of ter-

and the more interesting places it could go. And

Wilcox was happy to have Bickler work with gears and motors and come

up with the best rover design he

could.

Ruoff had already given Bickler a copy of a book by M. G. Bekker, the inventor of the Blue Rover. Wilcox also had a video of the original mobil-

SLRV from

of the

ity tests

JPL, which he

let

Bickler borrow. Bickler

Bickler liked to design

beginning to end.

What made It

was

driven,

the mid-sixties, after

flipped back

had been delivered to

went away

and analyze. But he did not

He

it

and

to study

like to

learn.

read books from

and forth through the Bekker book.

the Blue Rover so special?

a six- wheeled vehicle,

with

all

wheels independently

and divided into three bodies. The

front, middle,

electrically

and rear com-

partments were each supported by two wheels. The three compartments

were linked by

a spring-steel

compartments could the

rotate

two wheels. To make

member. For

on

their attach points at the center between

a right turn, the front

body would

wise, the rear counterclockwise; then the entire vehicle

ward

in

an arcing right turn.

and rear

steering, the front

When

rotate clock-

would

drive for-

the rover had traveled far

enough

"Almost As Good As a Bogie"

along

its

arc to

1?

be facing in the desired direction,

would

it

stop, straighten

the front and rear compartments, and drive forward again in a straight line. its

The

six- wheel

design

meant

that even

when one compartment with

two wheels was negotiating a hazard, the

vehicle

had four powered

still

wheels firmly on the ground providing traction and

stability. If

the front

wheels were handling an obstacle, the rear four were providing the

By the time

"push." rear

were the It

stabilizing influence,

and so on.

was magnificent! Nothing could touch

drive over rocks

be

the obstacle got to the middle wheels, the front and

like

one and one-half times

as

it.

high as

The Blue Rover could its

being able to drive your car over your dining

wheels. That

room

table!

the three bodies of the rover were connected to each other

each body could twist relative to the others. As a rain, all six

passage that intrigued him.

It

as a bogie."

that can pivot relative to each other. Bogies trucks,

and

ter-

it

relies

the vehicle

on

that

Blue

mean?

rigid linkages

were used on locomotives,

in military tanks.

a rock, the spring-steel linking the front

would be forced

to bend. Being a spring,

If

the front wheels

body it

onto the rock than they would

if

applying just their

That was the opposite of what you

really

own

to the rest of

tended to

bending, and that meant that the forward wheels pushed

this:

across a

like the

Now, what did

So there were limitations to the Blue Rover design.

began to climb

came

sprung suspension,

bogie suspension has no springs. Instead

mulled

a spring,

even in rough

result,

Bekker's book, Bickler

said that a

good

Rover's, could be "almost as

some

Because

wheels usually stayed on the ground.

One day while looking through

A

by

would

down

it.

Putting

more weight on

harder

weight. Bickler

wanted to

do. If

could take more weight off the wheels going over the obstacle,

be easier to get them past

resist

it

you

would

the wheels climbing

made negotiating the rock just that much more difficult. Another problem was that when more weight went to the front wheels, less went to the middle and rear wheels, giving them less traction. And traction on the rock

the other wheels

was

just

what was needed

to help get the front wheels

over the hump.

Bogies might be a terrain,

way

to allow the various wheels to

but without shifting weight to the wrong

conform

to the

wheels, like the Blue

SOJOURNER

18

Rover

did.

That must be what Bekker was getting

tence in Bekker's

book meant

bring to Bekker's party.

would need

all

that there

at.

Maybe

was something

one sen-

that

Bickler could

When rovers finally did get to other planets,

they

the mobility performance that could possibly be squeezed

out of a design. Bickler grinned. Bogies!

It-

Don Bickler grew up in Chicago, and went to college there to study mechanical engineering.

While

he became a co-op employee

in college,

He was

Stewart- Warner, a manufacturer of automotive equipment.

mediately put out on the factory

at

im-

Every week he would be moved

floor.

from one factory job to the next so he could be familiarized with each one.

He worked with the

mill operators,

sheet metal benders, then the welders, then the

and so on.

Early on, the foreman took Bickler aside and told him, "You Ve got to take your turn at the trash heap."

stock from the machinists.

The

The

trash

pieces, each

heap was

a pile of leftover

about ten feet long, almost

completely covered a workbench. They spilled over on the the workbench, and encroached

not be used, but scrap yard.

it

on nearby work

areas.

floor,

under

The stock could

was too long to be loaded into boxes and shipped to the

"What you've got to do,"

the foreman told Bickler,

"is

take this

hacksaw and saw each big piece into smaller pieces, then put the pieces the box so

we can send it out."

It

would probably take ten minutes

to

in

saw

way through one piece. That should keep him busy for a while. Bickler went to work. The first thing he noticed was that the stock's cross all

the

section

was not round, but hexagonal. That meant he could clamp

the vise and

it

wouldn't twist out. That

made

it

a lot easier to

it

in

work with.

One of Bickler's recent engineering classes had been "Strength of Materials." He thought about whether anything he had just learned would apply here. Well, since this was machining stock, it had to be ductile to make it easier to machine. If

it

was

ductile,

then

.

.

.

Bickler

clamped one piece

the vise, so that about eighteen inches worth stuck out one side.

grabbed the end and pulled

it

toward him, bending

it

around.

in

He

Then he

Asa

"Almost As Good

Bogie"

took the hacksaw, but instead of trying to saw near the point where

scored the steel

hand came away

the piece in his

19

all

the

way through, he just

came out of the

it

repeated the process on the next piece of stock.

He began

He seemed

He

to get into a

He

rhythm: Clamp. Bend. Score. Twist. Clamp. Bend. Score. Twist. this for a while.

and

vise, twisted,

clean, breaking off at the score point.

to be

making

did

a dent in the trash pile. After a

while longer, he had converted most of the pile into shippable lengths.

About

He

he noticed that the shop had gotten quiet.

this time,

and turned around. There were

stopped

fifteen machinists standing in a line

watching him. One of them shook his head. "We've had that trash pile twenty years," he

room, but then

it

said. "Sure,

Bickler could only muse,

applied

we

cut

it

down

a bit

when we run

overflows again. You just took care of "I

for

out of

at once/'

it all

guess engineers have their uses."

If

you

what you learned from books and professors, you could often find

solutions that eluded others. Building

perience of engineers

By the time

on the knowledge and

who had gone before him could be way through

Bickler had rotated his

all

a

codified ex-

magical thing.

of the positions,

he had earned the goodwill of the machinists and technicians on the tory floor. So

when

custom made

in

Bickler got his

sailboat ...

first

all

of

its

fittings

fac-

were

pure bronze and chrome-plated by the expert machinists

of Stewart- Warner.

As Bickler hunkered

down to figure

out

how a bogie suspension might beat

the Blue Rover in off-road performance, those days in Chicago were thirty

years behind him. His hair

He had

a

wry

was turning

smile that telegraphed

unfortunate engineer

who

silver,

What

tried to sell

thinking than true understanding. straightforward attitude that caused

He

though he

are

still

you trying

had

all

to pull:

of

to

it.

any

an idea based more on wishful still

had the young engineer's

him to argue passionately for the best

engineering solutions based solely on technical merit, having never

ac-

quired the political sensitivity with which most engineers were afflicted after

enough years

drive to

in business.

And he

still

possessed the same instinctive

uncover the better solution to the problem

in front

of him, and to

SOJOURNER

20

build

on the foundation provided by those who had examined

problems

in the past.

more and more

Bickler found himself

mobility design. attended.

working

gram

Ken Waldron, Ohio

at

involved in the area of rover

He heard about a meeting that sounded interesting, so he noted researcher in the

a

State University,

field

of robotic vehicles,

had been funded by one of JPL's pro-

study six-wheeled vehicles and their potential perfor-

offices to

mance. Waldron was sat in

similar

on the review

visiting JPL to present his plans for review. Bickler

what he might

to see

When

learn.

he didn't under-

stand something, he asked questions. Soon after the review, he was asked to be the JPL contract

monitor

know something after all And then he got invited .

meeting

.

for the

to another meeting.

was an organizing

It

was doing

workshop was experts

on

MRSR

workshop on "Mars Rover Sample Return" or

for a

in that

did

.

possible to bring samples of rock I

Maybe he

State contract.

which would examine the various technologies

short,

what

Ohio

and

soil

room." Soon

to bring together

it

that

back from Mars.

became

clear.

would make didn't

"I

The

from around the country

for

as

it

know

intent of the

many of the

rover-related technologies as possible. Brian Wilcox

was

al-

ready going to be leading a session on local navigation and hazard avoidance.

He wanted

Bickler to chair the session

Bickler protested:

He

on rover mobility concepts.

know who

didn't even

the vehicle design gurus

were. Wilcox persisted: "Don't worry about that.

and numbers of

And call

that's

all

what Bickler

up, he says,

You just have

the right people. did.

"So

I

get

give

you the names

to pick

up the phone."

I'll

on the phone, and the first guy

Tm not coming unless so-and-so's coming.' So

next guy, and he says he's not coming unless the ing back and forth, and finally get a pretty chairing this session.

first

this

That wasn't entirely true. At just about the

call

together.

I

up the

guy's there. So I'm

good group

And that's how I got into

I

call-

So I end up

mobility business."

same

time, Bickler gener-

He made his first model out of plywood. The six wheels were the leftovers you got when you used a hole ated his

first

saw to put

bogie-based mobility concept.

a hole in a

door to

install a

doorknob. The model was simple,

without motors. Just wheels and wooden linkages between them, so you could push

it

around on

a tabletop.

But even

as simple as

it

was, you could

"Almost As Good As a Bogie"

see that

it

could

ble in

on

over a block of

You could

tipping over. pivots,

roll

wood higher than its

all six

terrain. Unlike the

wheels, even

when

at

all;

let

that

limits of the

meant it would be

sta-

Blue Rover, the model kept equal weight

climbing over an obstacle. The only problem

with the design was that you couldn't steer ing pivots to

wheels without

any wheel off the ground, to the

lift

without any other wheel moving

rough

2

it.

the wheels turn, the linkages

you

If

tried to install steer-

would ruin the ground clear-

ance of the vehicle, so you really couldn't deal with the rough terrain you originally

thought you could.

A year new

later Bickler

that problem, at least

design incorporated "virtual" pivots. Using a

linkages, the all

had solved

new design acted just like

the old

paper.

number of

wooden model,

The

four-bar

except that

the links sat above the wheels, at a height that didn't interfere with

ground clearance. Only one

To make each wheel of those thin

links.

steer,

One

the four-bar linkages

People started calling the

So

far,

Bickler only

to build one, but he

you just had

to put a rotational joint into each

who looked

engineer

at

any desired

size

his sketches

wanted

this

one

at Bickler's

by tracing

it.

The name

stuck:

He was ready He would need

of the pantograph.

to

be motorized.

there were

more

de-

going on to define possible future Mars rover missions.

Donna Shirley, who was leading the MRSR study team, funding to continue looking build anything. Bickler as well as his

design said

"Bickler pantograph."

money from somewhere for parts. By now, mid- 1988, tailed studies

new

a "pantograph," a drafting device

new design the

had

down to the axis of each wheel.

thin link went

reminded him of

used to copy a drawing

at rover

wanted

space' of options

gave Bickler some

money to pantograph would move

mobility concepts, but no

to prove that the it

would. But Shirley told him no: "We're

this

study We're only examining the 'trade-

paper analysis said

not building any hardware for

He

on

and developing a conceptual design." Bickler was stymied.

could not misdirect Shirley's funds to do something she specifically

forbade; but endless evaluation of "tradespaces"

was

just not his idea of

engineering.

One

day, Shirley

was presenting the

results

of the rover study. The

audience was a review board composed of high-ranking JPL and representatives.

One

of the slides Shirley

showed was

a sketch

NASA of the

SOJOURNER

22

pantograph.

A

review board

member

stopped her and said the panto-

graph seemed overcomplicated to him. "What's the use of plexity?"

and he

he wanted to

dutifully

mance of

The

in

Bickler

was

sitting in the

that

com-

back of the room,

with an explanation of the mobility perfor-

The reviewer was not convinced. So Bickler asked, see a model of this design that you could run around

the design.

"Would you on the

know

chimed

all

like to

Then you could really

tabletop?

see

what the pantograph can

do."

reviewer, along with the rest of the board, leapt at the offer.

had what he wanted. The review board outranked

Bickler

had just been given an endorsement

figured that he

to build a

Shirley.

He

pantograph

model.

With could at

scant funding, Bickler built as

home

in his garage

much

of the pantograph as he

workshop. Instead of complete blueprints

generated by the JPL design room, he gave the JPL machine shop SV2-X11 -inch sheets

of paper with drawings hand done with a

And the few parts he

felt-tip

marker.

asked the shop to produce were mostly not finished

components, but merely aluminum stock cut to appropriate lengths.

bought

motors

electric

to construct fiberglass

shaped so

when

at the local surplus store. Bickler

"dome" wheels

on hard

that,

in his garage.

surfaces, the rover

the vehicle got into soft sand,

it

would

would

The wheels were

ride

on the

rims, but

More of the ground, providing more

start to sink.

wheel would then come into contact with the surface area

He

even managed

and helping to prevent further sinkage.

It

was

like the differ-

ence between walking in snow in boots or in snowshoes: The person in boots might sink in up to his neck, while the guy in snowshoes could walk along nearly on the surface.

At plete

least, that

was the

idea. Bickler

pantograph and give

about two

feet long,

tor in each wheel. to a battery

backward.

and would

and

a

a

try.

Once put

to assemble the

together, the

com-

model was

and had five-inch-diameter wheels. There was

a

mo-

The motors were ganged together, and wires led back switch, so the rover model could be driven forward or

When he slip

it

would have

first

on hard

over each wheel to give Bickler's analysis

tried out the vehicle, the

wheels were too

slick,

floors or tabletops. Bickler slipped rubber bands

them more

traction.

had indicated that the pantograph should be able

to

"Almost As Good As a Bogie"

climb steps one and a half times the height of Rover. So he had the carpenter shop

low the pantograph

was

make him

to demonstrate just

what it was capable

Between phone

calls

just wasn't getting a chance to try out his

quieted down,

it

was

after five o'clock.

and placed the six-wheeled model the step with ease!

wanted

to

slippage,

it.

of.

and

al-

When the

Now

he'd see

do!

Well, soon, anyway.

into the hall.

a carpeted step the appro-

he went over to the shop and grabbed

ready,

what the design could

Bickler

wheels, just like the Blue

its

The carpeting would minimize any wheel

priate height.

step

23

and other interruptions, he

new vehicle. By the

time things

He pushed the step against the wall

in front of

it.

The pantograph went up

could climb steps as high as the Blue Rover could.

It

show someone what he'd accomplished. He ran out

Most everyone had already gone home. Across the

the division manager's office, and he

was

still in.

played with the pantograph, driving up and

So

for the next

down the

various locations, steering the wheels, like

two

step, trying

kids with a

hall

was

hour they it

new

out in

toy on

Christmas Day.

With

a

few more experiments Bickler was able to prove that the pan-

tograph could cross crevasses as wide as 40 percent of the length of the

And

rover.

the pantograph design gave

higher than the Blue Rover, so six

it

it

ground clearance three times

could go over

much rougher terrain.

wheels could be steered, so the pantograph could either turn

Blue Rover, or instead turn

all

wheels in the same direction,

All

like the

and move

off

to the side like a crab.

* After playing with the

pantograph some more, Bickler discovered a draw-

back in the design:

had

step,

it

had no

initial level.

like a

It

a

problem with "bumps."

trouble; but the front wheels never

On

the other hand,

if

bump

the rover

came

dropped down to

to a

their

the pantograph drove over a bump)

—the front wheels ended up

lone rock

If

at the

same

level after

going

The middle wheels would then have greater difficulty crossing the same bump, and with large enough bumps would not be able to get over the bumps at all. Bickler was disappointed. Why would over the

as before.

the pantograph design be worse than the Blue Rover in this particular

SOJOURNER

24

way? Surprise! The Blue Rover had the same problem! Bekker must have

known

this,

but never mentioned

it

in his book.

And if you knew what to

SLRV getting stuck on rocks in the videotape, then backing off and going another way Bekker had kept a secret! look

for,

you could

see the

The design of the high-mobility rover chassis did not end with the panto-

graph, or "Bicklermobile" as

it

was sometimes

also called.

The

MRSR re-

view board's objection to the pantograph had not gone away: The pantograph was complex.

and

risk. It

Mars.

If

rocket

a lot of parts.

Those parts added weight

pound just to launch

a payload to

the total mass of the payload were too high, then the launch

is

its

able to reach

pushed to keep the weight of

tem is given given

had

cost thousands of dollars per

would simply not be

mission

is

It

a

mass

allocation,

its

its

target.

So every deep space

spacecraft low. Every subsys-

and every assembly within each subsystem

own fraction of that allocation. And a heavy rover would mean

that fewer science instruments could be flown to Mars.

With deep space robotic missions

there

is

no one

to repair

failures during the mission. While clever engineers can, at

sign

equipment

some

cost, de-

backup systems, one of the best and cheapest ways to make equip-

make it as simple and elegant as possible while still getting the job done. The multiple four-bar linkages of the pantograph just looked like a mechanism that could jam or break, begging the quesment

reliable

is

to

tion "Is there a simpler way?"

The answer

to that question

would be

called "Rocky."

THREE

0FF-R0AD1NG WITH NO ONE AT THE WHEEL

white paint

The the

was

ratty,

mottled, and cracked.

The rubber covering

spring-wire-loop wheels had rotted through, collapsing the

wheels and creating the appearance of some type of mechanical beast prone in

its

black nest.

The

last

time Brian Wilcox had seen the Sur-

veyor Lunar Roving Vehicle prototype, tion.

it

had been

But that had been over twenty years before.

much better condiHe had been twelve

in

years old.

In

the late 1950s Brian

a classified

program

that launched rockets craft

s

father,

Howard Wilcox, was program manager for

at the Naval Ordnance Test Station at China Lake

from

a jet fighter.

Launching from

a high-flying air-

above the thickest part of the atmosphere allowed a smaller rocket to

put a payload into launches placed at the time,

orbit,

compared

satellites in

to a ground-launched missile.

These

Earth orbit in mid- 1958 and, although secret

represented the fourth and

fifth

successful U.S. satellites,

within a year of the Russian-launched Sputnik.

Howard Wilcox had moved on to become head of Research and Engineering at General Motors. The space program was just getting under way. In addition to the manned Mercury program, robotic probes By

1960,

SOJOURNER

26

were being designed to

Moon.

travel to the

In 1962,

JPL issued

a

Request

For Proposal for a high-mobility vehicle intended to operate on the sur-

Moon. Wilcox had recently hired a sharp mechanical engineer,

face of the

M. G. "Greg" Bekker, who had tion

a flair for issues of vehicle-terrain interac-

and mobility design. Wilcox turned to Bekker's team to respond to

the announcement.

Bekker was up to the

team

With internal company funding, he and his

task.

built a six-wheeled vehicle

looked

like

it

was

all

wheels.

with balloon

They made

a

tires that at first

glance

movie of the vehicle driving

through a daunting desert obstacle course with lava flows and boulders.

With

its

melodramatic musical background, future robotics engineers

JPL would

over almost anything. General Motors ate,

and

Moon. However, the

refer to the film as Rover Gladiators on the

mobility capability of the vehicle was impressive.

deliver

won

looked like

It

it

could go

the contract to build, evalu-

two prototype Surveyor Lunar Rover Vehicles (SLRVs)

JPL by 1965. The SLRV prototypes were to be smaller than the hicle.

They needed

to be able to

fit

at

to

original concept ve-

into the already-designed Surveyor lu-

nar lander. Another requirement imposed on the prototypes was low weight, so that they the Surveyor, and to operate.

would not exceed the payload carrying

would require

less

of the scarce onboard battery power

The vehicles were each given a single

mote operator

capacity of

TV camera to allow a reand so

to see the terrain ahead of the rover

safely drive

it

across simulated lunar terrains. In 1964 there

was an open house

facility.

Among

rovers.

The SLRVs had not

open house,

at the

General Motors Santa Barbara

other displays and demonstrations was one about lunar

largely family

been delivered to JPL, so

yet

and

were given the opportunity to

friends of General

drive

one of the SLRVs on

could be seen only through the rover's

own

visitors to the

Motors employees, a test track that

camera. The track was a

rugged "moonscape" hidden on the other side of the building from where the

open house demonstration was

set up.

Brian Wilcox, barely out of elementary school, drove the rover that his father's

team had

ward or backward

built.

The

control box let

a fixed distance, stop,

him

and turn

drive the vehicle for-

left

or right. Adjusting

Off-roading with No One at the Wheel

SLRV into

the controls, he could put the

a shallow,

2?

medium, or sharp turn

Then, by hitting the "forward" button, he would make the rover

angle.

through the arc defined by the steering angle. By

drive four feet or so

watching the images from the onboard camera, Wilcox found keep the

SLRV between

the small rocks that defined the

it

easy to

bounds of the

"safe" path.

mode

But there was another the rover:

They delayed

to simulate the time

coming the

all

way from

the

SLRV was

up

for driving

the display of the video images by a few seconds

would

it

the engineers had rigged

the

take for those images to arrive

Moon, which was where

intended to go. The

Moon was 240,000

traveling at the speed of light, the pictures

they were

if

the final version of

away

miles

so even

from the rover would take

about a second and a quarter to reach the Earth, or two and a half seconds for a round- trip message.

Even worse

signers set the video feed to shut off

This also was

when

it

came

its

to a stop. If

behind the wheel of

like sitting

would-be

drivers, the

whenever the rover was

because a rover driving across the

realistic,

be able to keep the antenna of except

for

a car,

de-

in motion.

Moon wouldn't

video transmitter pointed

you drove the SLRV

demo

in this

at the

mode,

Earth it

was

looking where you wanted to go,

and then closing your eyes while you put your foot on the

gas; after

you

braked, you opened your eyes again to see where you had ended up.

When Wilcox tried to

drive the

SLRV in its

simulated lunar mode, he

got into trouble. Without visual feedback from the rover as

he couldn't

tell

exactly

couldn't see the rover

had no cues

to

how

itself,

the rover got to

remind him whether the rover was

apparently directly in front. But set distance ahead, the

The

ended

in the

up. Since

he

middle of a turn

The video image might show

when he commanded

none of the rocks he'd

last

a clear

path

the rover to drive

next image that appeared would

pletely different view, with sight.

it

was moving,

but only the terrain in front of the rover, he

or aligned facing straight ahead.

its

where

it

show

a

com-

seen anywhere

in

rover had actually turned ninety degrees to the right, and was

looking off the edge of the safe path into treacherous terrain. But the video information wasn't sufficient to tually

tions

tell

him

was what had

ac-

commanded mohe wouldn't know whether he had

happened. Unless he could remember

and turns he had used previously,

that that

all

of the

SOJOURNER

28

Which choice should he make to correct the problem? He wasn't sure. So he guessed, and often guessed wrong. The attempted corrections could become wilder and wilder, until Brian (and everyone turned left or

who

else

right.

and time-delayed rover)

tried to drive the video-impaired

to keep the

SLRV within its

test track.

Of

course,

the continuous video feed back on, driving the

if

failed

the engineers turned

SLRV

suddenly got easy

again.

Brian Wilcox learned a lesson that day. ing

him huge amounts of

fore.

Without

Soon

useful information he hadn't appreciated be-

that feedback,

after that

The continuous video was giv-

you just couldn't

drive.

open house, the General Motors SLRV was delivered

toJPL.

Howard Wilcox was trained

and he began teaching his son the

as a physicist,

science early on. Brian took to the subject.

"I

was good

at

it

because

I

was

And I found an amazing power that came from being able to analyze things." He majored in physics and mathematics at UC Santa Bar-

tutored in

it.

bara, attaining dual degrees,

and was accepted into the

UC Berkeley grad-

uate physics program, the premier such program in the country. But after the

first

term

at Berkeley, he'd

had enough. "You can do

with undergraduate physics. But graduate

back to

a

level physics ..." Brian

company where he had worked during companies of his own. One of

started a couple of

morphic Systems, produced one of the puter hobbyist, the Poly-88. units per

released It

month

its

in 1977.

first

The company

Then another

practical things

college,

went

and eventually

those companies, Poly-

microcomputers

for the

com-

did well, selling hundreds of

startup,

named Apple Computer,

own computer, and Polymorphic Systems was doomed.

was time

to find

ested in robotics for botics business.

computers, so

it

something

some

else to do. Brian

time, and he decided he

"My background was seemed

tems was the place that

physics,

Wilcox had been

wanted

my

inter-

to be in the ro-

application

had been

natural that computer control of mechanical sys-

my

interests

were best used,

in particular micro-

processor systems." So he began looking for opportunities in robotics.

Helpful information came from a surprising source: his father.

Howard

Off-roading with No One at the Wheel

Wilcox had worked on

a

number of projects

29

since his days at

GM. One

of

the recent projects had been a survey of people and institutions doing re-

search and development in the field of robotics.

Howard Wilcox

gave his

son a copy of the study, and the younger Wilcox sent out resumes.

One

of

those resumes went to Carl RuofF at the Jet Propulsion Laboratory.

The day Brian Wilcox's resume reached been told by people to

Once the GM

SLRV was

delivered to JPL in the mid-1960s, engineers there

performance

its

to send a robotic rover to the

fizzled out.

entrepreneurial"

Brian was just what he needed.

drove the rover, evaluating

program

more "well-educated but

his boss to find

hire.

Ruoff's hands, Ruoff had just

The SLRV went

in the

Arroyo Seco. But the

Moon before

into storage.

the Apollo landings

Sometime

in 1972,

Howard

Primus, a technician in the Automation and Control section, took possession of the rover.

No one else seemed to think the rover was good for any-

thing, except taking rover, Primus's

up storage

managers would

Whenever they came across the him to have it hauled away. Instead,

space. tell

Howard would move the SLRV to a different storage place, keeping the rover out of sight as much as possible. Months would go by, then a supervisor would stumble across the SLRV "Get rid of that thing!" he would say. Primus would just move the rover again. He just couldn't see disposing of such a unique machine; surely, someday it would find a new purpose. "Howard just loved that rover," Carl Ruoff would later remark. Howard Primus's shell game went on for about ten years. By 1982, Ruoff was supervisor of the Robotics group. One day Ruoff got a call telling him that the storage trailers outside of his research laboratory needed to be removed. The trailers contained overflow materials from the teleoperations lab, so that made them Ruoff's responsibility. They sat in a narrow outdoor parking vision

manager wanted

key to the

trailers

south of Building 198, and the

di-

that space for additional parking. So Ruoff got the

and went to see just what was stored

opened up one of the

went

lot just to the

trailers,

to get Brian [Wilcox]

here. Let's take a look at

and "By God,

and

it.'

I

them.

there's this white rover.

said, 'Hey, Brian,

So Brian and

in

I

He So

I

think we've got a rover

went and opened

this trailer

SOJOURNER

30

and he

[again],

says,

'You know,

twelve years old, and

I

drove

I

think that's the rover

into a ditch.

it

We

drove

I

when I was

can't really get rid of this

SLRV to working order and restart He had Howard Primus bring the vehicle

rover/" Ruoff wanted to restore the robotic vehicle research at JPL. into the

basement laboratory

were simple: Make

it

The

U.S.

RuofFs instructions to Brian Wilcox

work again.

knew of

Ruoff needed funding and cles.

area.

possible military uses for robotic vehi-

Army was getting interested in the

idea of a robotic recon-

naissance vehicle, one that could go places where soldiers were unlikely to survive.

Ruoff went to the JPL program

office representative for military-

funded programs and asked for money to make the SLRV operational.

The manager gave him Ruoff, Wilcox,

The

thousand

dollars.

and Primus went to work. Surprisingly enough, once

they had the go-ahead, tioning.

five

it

needed replacement rechargeable

vehicle

wheels. Primus had carefully hoarded spare parts for

SLRV had been good enough

otherwise ignored.

until a

new

The wheels were another

set

many of the

all

of those years the

A spare battery set was among them,

could be identified, ordered, and

installed.

The wire-mesh, rubber-covered wheels

matter:

had been an expensive specialized design possess

SLRV rover funcbatteries and new

took only a few days to get the

in the early 1960s, intended to

characteristics of the final designed-for-flight wheels.

(Although the rubber covering would not have survived the temperature variations of the lunar surface.)

lightweight,

uneven

The wire-mesh wheels had been very

which was important

terrains. If the vehicle

in enabling the

were too heavy,

move under its own power. There was no way

it

machine

to

perform

just wouldn't

in

be able to

they could afford to dupli-

cate the flightlike wheels.

But then again, they didn't have

do with the SLRV was to create search,

a

to.

What Ruoff and Wilcox wanted to

working platform

for robotic vehicle re-

one that would allow them to experiment with control

and maybe onboard robotic

"intelligence." If the rover could

sonably well, that would be good enough. Vehicle wheels

would do the

trick.

Maybe

strategies

move

rea-

plain old All-Terrain-

So Ruoff and Primus drove over to the

Off-roading with No One at the Wheel

3

Honda motorcycle dealer in Pasadena and took a look at the ATVs there. Some of the wheels looked like just about the right size. But the wheels were

steel

and seemed to be too heavy for the motors on the SLRV to han-

dle.

Next, Primus went to a dune

that

would do the job. Once he had the wheels

adapters to

six

mount

a

buggy

and found some wheels

store,

Primus machined

in hand,

wheel hub to each of the

rover's drive motors.

To

reduce the weight of the wheels further, Ruoff asked Primus to grind the treads off the

ing

its

little,

tires,

making

the rubber as thin as possible while maintain-

structural integrity. Since the

SLRV drove so slowly and weighed so

the tires didn't really even need to be inflated. At a top speed of half

a mile or so per hour, there

was

little

risk

of the

tires falling off

the wheel

rims.

When Wilcox had

first

come to work at JPL, he had been assigned to Ruoff 's

Army

study team that was jointly funded by the U.S.

Engineer Topo-

graphic Laboratories (ETL) and the Defense Advanced Research Projects

Agency (DARPA). The purpose of the study was

to identify the research

necessary to create robotic reconnaissance vehicles useful to the military.

The team focused on how

to direct a rover through

roundings and avoid the hazards in to

its

path.

its

immediate

The long-term

sur-

was

objective

human interIt finds its own way there, even

produce autonomous vehicles that required almost no

vention. Give the robot a goal, miles away.

avoiding the

enemy if necessary.

to intelligently

command

In the short term,

a teleoperated vehicle

coming up with

was

a

more

a

way

realistic

ob-

jective.

The

challenges of directing a robotic vehicle

on the

were

battlefield

not so different from operating a rover exploring another planet. You couldn't afford to

was no way

communicate with

to give a rover

a rover

enough power

to send a strong

the millions of miles back to Earth for continuous video.

it

and destroy the robot. So

in

the vehicle needed to be able to do

Now

that they

had the SLRV

it

both

There

enough

signal

If a battlefield re-

connaissance robot sent a strong signal too frequently, the track

often:

on Mars very

cases,

enemy could

whoever was operating

with very limited information.

rover,

Wilcox and JPL were

in a better

SOJOURNER

32

position to market robotic vehicle technologies to the US. military.

SLRV was

gies without starting

he had

on which

a platform

from

implement and prove those technolo-

to

scratch.

envisioned during the

first

The

And Wilcox had

ETL

a navigation concept

one he

study,

way-

called "stereo

point designation."

you

If

watch video coming back from a camera

can't afford to

mounted on

what can you do? You

the rover,

was around the treme from a

what the

rover,

live

needed to know what

still

terrain looked like. Well, the other ex-

TV picture would be just a single frozen picture or one

video frame. But that wasn't quite enough. With just one picture, a close-

up pebble might take on the appearance of terrains

would appear

flat in a

more

a

distant boulder.

Some

photograph, but be obviously undulating

to a person standing next to the rover.

What you needed was two

eyes

the rover: stereo cameras to provide a 3-D view of the terrain.

predatory creatures with two eyes

on

Most

—humans included—use binocular

vi-

sion to estimate distance to objects in the environment around them.

With two cameras viewing the same

object,

the range of the object by triangulation.

of the two camera lenses

is

it

The

was

line

possible to determine

connecting the centers

the base of the triangle; the lines between the

cameras and the object form the other two

sides of the triangle. If the

cameras were mounted side-by-side a fixed distance

apart, the

known

geometry would enable range determination. People were also natural experts at doing something that

gramming terrain

a

computer

no one

yet

to do: looking into a

had

a very

good way of

pro-

3-D display of a rock-strewn

and sensing immediately which areas were

safe

and which were to

be avoided. Wilcox's insight was that a stereo vision display could also be used as a

command

joystick

move

input device.

If

he could design a 3-D cursor, operated by a

and displayed in the middle of the 3-D

the cursor until

it

terrain,

appeared to be located

at a

then a person could

good

safe target loca-

tion for the rover. If a person designated several of these target points in

the display, they

would

constitute a safe path. If the system

was properly

calibrated, the target points in the display could automatically

verted to target coordinates, and then into a series of motion for the vehicle.

The commands could then be transmitted

be con-

commands

to the rover,

Off-roading with No One at the Wheel

which would then execute them, driving

—

a safe path

33

through the interme-

—

way waypoints to the final destination. Once the commanded destination, it would take a new pair of stereo images and send them back to the operator. Then the whole process diate points along the

rover reached

its

would be repeated until the rover achieved its ultimate Wilcox proposed

concept as a

this

JPL could provide, one

that

SLRV

In mid- 1984,

new

destination.

vehicle control technology

that could be tested out using the restored

Wilcox and Joe Hanson, one of the JPL

points-of-

Army customers, made a marketing trip to the US. Army Tank Automotive Command (TACOM) to try to sell the idea, with contact to potential

no immediate

success.

On the plane flight home, Wilcox and Hanson dis-

cussed what they would need to do to get

was

that

you almost had

vinced that

it

was

to have a

doable.

TACOM to bite.

working version

Hanson thought it was

first.

The problem

Wilcox was con-

a bigger job than

Wilcox

imagined. While Wilcox and Hanson talked, they had a couple of drinks. Before the airplane landed in Los Angeles, Wilcox boasted that he could

up and running on

personally get stereo-image -based target designation the

SLRV working

only a couple of weeks of evenings and weekends.

The implementation would be it

would be

Hanson

First

crude, rough, and inaccurate, but at least

a concept demonstration. didn't believe

Wilcox needed to

mount

stereo to work, the cameras ate distance apart.

a pair of video

camera's view. Otherwise,

it

would be

Ideally,

side

by

side

For

an appropri-

vertical position in each

person with one eye looking floor.

The

spare cameras field

each camera would have an optical mount that would

justed, enabling the pair of

(pan,

tilt,

roll) to

al-

be independently ad-

cameras to be carefully aligned with each

Wilcox didn't have the money or the time to buy or build

mount, so he

SLRV

had lenses that gave them a thirty-degree

low each type of camera rotation

other.

same

like a

right.

aligned so that an object

and the other pointed toward the

available in the robotics lab

of view.

at the

almost

cameras on the

had to be mounted

The cameras should be

viewed by both cameras would be

at the ceiling

He was

could be done.

it

drilled holes in a simple

a

complex

aluminum bar and mounted

the

34

SOJOURNER

cameras about twenty inches

apart.

He

then affixed the bar to the top of

compartment of the SLRV, so

the front

that the cameras faced forward.

Now Wilcox had to get his hands on a stereo vision display. He knew that the teleoperations

some of

displays in

group

their

system was not as impressive

frame

steel

which two

in

one

to each other,

tor above

as

first,

mirror was mounted

result

would be

but facing

TV

had

down toward

and

itself to display

a right,

ter.

The

"left"

in the 1950s.

polarization

matched the ferent

polarizing

left

see

TV could be seen,

Of

course, the mirror-image of correct for

this,

on the

one wire

CRT

and

the video feed backwards. In this

display effect, a person

worn by movie filters in

would don

on the

left side

a

audiences for 3-D

the glasses

would allow

had been polarized in the same direction

"right" monitor.

TV monitor. her

would

fil-

as the

fil-

of the glasses matched that of the

in the stereo display, while the right side of the glasses

ing into the display his or

The

light that

monitor

a polarizing

right-side-right.

pair of polarized glasses, like those

through only

first

A

the floor.

angle between

and the reflected backward image was

To achieve the three-dimensional movies

was

to be resoldered, reversing the scan

two wrongs did make

left-side-left

a

the half-silvered mirror; the other screen

was just that: backwards. To

causing the monitor case,

was mostly just

that a person looking into the display

visible as a reflection in the mirror.

inside the

borrow one. The

at a forty-five-degree

faced. In front of each screen

somewhat dimmed, through the second monitor

It

person viewing, the other moni-

the images from both monitors superimposed: the

was

to

one might imagine.

at eye-level facing the

where the two monitors

The

basement of Building 198 used 3-D

TV monitors had been mounted at right angles

and forward of the

half-silvered

ter.

in the

work and got permission

So each eye would see an image from a

dif-

Properly hooked up, this meant that the person look-

would

see the

image from the

left

SLRV camera with

eye and the image from the right camera with his or her

right eye. This created a sense of depth

and understanding of the scene

ahead of the rover that no single image could have supplied.

A

real stereo designation

tronics that

would capture

system would have "frame grabbers,"

a single video frame

elec-

from each of the video

Off-roadingwrthNoOneattheWhee

cameras on board the era and one

time. But Wilcox

ran

fifty feet

of coax video cables back from the

substitute for the frozen

still

back

sen:

first

images that the

a tool for telling the rover

means of moving

demonstration,

in the lab that

they

around

it

had

what

live

It

looked

vertical bars

cam-

drawn

video would

was

would a

He

rover saw.

give

him

a

wav

a

to

custom image processor

counted the rows and columns of images

make

like a

directly into

3-D crosshairs and

the camera. Wilcox designed and built

ification to the processor to

and

to do: a

in the display. This

circuitry* that

came out of

stereo display

left

system would depend on.

real

designate points in the stereo image. There

as

SLRY cameras

now had a 3-D viewer that showed him what the

needed

the

:o the control station at a

had no computer or frame grabbers on the SLRY So he

the stereo display. For purposes of the

Wilcox

two images one from

rover, so that only

from the nght would be

35

big

a

a

stereoscopic crosshairs overlay

white -

ed-

:

on the

sign constructed of horizontal

two camera images. The

into the

simple n:

the flexibility to be shifted not only left-right and

crosshairs

up-down

had

in the display

but also rn-out. representing the distance out in front of the rover. To do

and right-eye

that required separate left-eye

zontal bars of the

two

crosshairs

other, the vertical bars

would

viewer. Wilcox .After

would be

offset

the crosshairs

bought

a joystick

While the

from one another. This

create a perception of depth to the

would make

offset

crosshairs:

hori-

would always be superimposed on each

move

human

directly

from the

offset

viewer. Adjusting the

away from or toward

local electronics surplus house.

modifying and interfacing the joystick to the image processing

tem, he could control

all

software, the

3-D crosshairs position on the screen into

which the rover could be directed

computer could generate output

of the eight motors

SLRY Wilcox

ran

sys-

three directions of crosshairs motion. Appropri-

ate software could convert the

target location to

the

With more on any or

all

motors and two steering motors on the

six drive

more wires

to drive.

signals to turn

a

to earn" the output signals

puter to newly installed analog amplifiers on the

from the com-

SLRY Now

for the first

time he had an off-board computer capable of driving the rover.

Wilcox went to the JPL materials supply colors

were

available.

He

called

m search of paint. Only a tew

one of the JPL in-house video prodm

SOJOURNER

36

and asked her which choice would look best to the camera. The answer

came

among

back: blue. So Wilcox selected the only blue paint from

options and gave the

SLRV its

first

new coat of paint in twenty years.

Wilcox was running out of time. As he worked to develop the

and weekends" than he had

first

more of

version of stereo target designation, he found himself devoting his "evenings

his

imagined when he

originally

had made the bet with Hanson. Wilcox's

first

implementation of

crude. There were

had

no sensors on the motors

actually driven or turned.

rover's

ond. So

if

the target location were

the motors for

two seconds.

would run the motors rover

was more

for

variable:

It

to

measure

The computer program

motors just assumed that the

If

was very

a rover driving algorithm

how far the rover that operated the

speed was one foot per

rover's

sec-

two feet ahead, the software would run

the destination were twenty feet away,

about twenty seconds, and so on. But the

it

real

would slow down when going uphill, speed up

on the downhill; and the wheels could

slip

on sandy

surfaces, so that the

would ideally. The

steering

motors on

Wilcox's rover were run by timing just like the wheels. To get a

maximum

vehicle covered less distance than

turn, stops.

it

you just ran the motors long enough Going straight was harder,

the right

since

amount of time away from

it

to be sure they

the turn limit.

turn was different for front and back, and for

major complication that Wilcox's "quick and tended to handle. Just cle

forward

as the drive

at a certain distance

had

involved running each

left

and

The time right.

hit their

motor just it

took to

That was one

dirty" software

motors were assumed to

was not

roll

in-

the vehi-

per second, so the steering motors were

expected to turn the front and rear vehicle compartments through a certain angle per unit time. it

would

that as

it

When the rover completed its commanded turn,

usually be pointed only in the general direction of

drove forward

one

target, so

its

side of

intended

it

drifted significantly to

it

worked! Wilcox could look into the stereo

its

destination.

But, after a fashion, play, adjust

choose a

the 3-D crosshairs using the joystick, and

target. If

he wanted

to,

he could

hit the

dis-

button to

select several targets, creating a

multi-segment path, but he would then have to remember where the earlier targets

were, since he didn't have any

way

to display

them on

the

Off-roading with No One at the Wheel

screen.

That would come

later.

3?

He'd done what he'd promised on the

air-

plane.

When Joe Hanson saw the system,

Wilcox's first effort did not stration,

work

well

he was floored.

enough

to ever give a live

demon-

but the basic functionality was there. Enough that the rover and

the control station could be used to explain the basic elements of the sys-

tem. Wilcox process:

made

a

homegrown video that went through each step

The rover transmits

operator designates a target position in those images; to the vehicle;

and the rover goes

further discussions,

of the

stereo pictures back to the control station; the

there.

On

commands

are sent

the strength of that tape and

TACOM funded a new JPL task starting in late

1984 to

develop and demonstrate a stereo-waypoint designation system.

The new system

filled a

niche between a remotely driven or teleoper-

ated vehicle and the holy grail of the fully

needed no guidance

that

plete

its

navigation task.

driven, with the

at all

A

from human operators

teleoperated rover

up from

vehicle

in order to

com-

was just being remotely

onboard communications system allowing the controls

and human driver to be located somewhere step

autonomous robotic

this.

Charley Beaudette, the

else.

The new system was

TACOM

a

technical director for

new term: Computer- Aided Remote Driving, or CARD. years later, when Sojourner rolled across the dust of toward rocks named after cartoon characters, it would be a

the task, coined a

More than twelve Ares

Vallis

variation of

CARD that would guide her.

But there was yet a long path to follow to prove technology.

CARD

a practical

FOUR

THE RIGHT PLACE AT THE RIGHT TIME

two projects Ed Kan and Neville Marzwell had hired

The were coming to an end. was

I

successfully completing

question remained: I

was learning Those

ends.

didn't have

What was

that projects

I

to

do

my JPL assignments, yet a nagging

going to work on next?

and tasks

me

at JPL

I

started to

worry

had beginnings, middles, and

tasks provided thirteen-digit charge

numbers.

And

if

you

an authorized charge number to put on your time card, you

could be out of a job.

When Ed Kan first offered me the job at JPL, he had

been noncommittal about future employment, only indicating that it was likely that It

other things would

come

up,

if

I

did well.

turned out that JPL had a special charge number for employees

had no of the

real tasks to charge to. artificial

It

was referred to as "090" for the

organization within JPL that administered

was out of work at JPL, they were "on oh-nine-oh"

it.

If

who

number

someone

until they either

found

work or were laid off. During the time you were charging to 090, your job was to find a job. JPL was generous: Each employee was entitled to one month plus one week of 090 charging for every year he or she had worked at JPL.

No one wanted to get on oh-nine-oh. I

was

still

a JPL rookie,

and

I

wondered where

my next job was com-

The Right Place

at the Right

Time

39

One

ing from. Yet opportunities did present themselves.

door was the task manager for

in the building next

experiment.

The

intent of the experiment

was

a

of the engineers

Space Shuttle

and

to build

flight

test a force-

torque sensor to be installed on the shuttle's manipulator arm.

I

was

as-

tonished to learn that the huge Space Shuttle manipulator, regularly used to

lift

large satellites out of the cargo

when

bay and deploy them into

orbit,

had

made contact with an object. For the astronauts controlling the Remote Manipulator System, it was like trying to pick up delicate pieces of machinery with an arm shot full of Novocain. The only protection against knocking the mechanical arm against the side no sensors on

it

to

tell

it

of the shuttle was the vigilance of the astronauts as they watched through

windows or TV cameras looking into the cargo bay. The sor

was

to

be mounted

at the "wrist"

force-torque sen-

of the manipulator.

It

would com-

municate back to the astronaut operating the arm the direction and

magnitude of the forces the arm was generating in the cargo bay.

broken and ager asked the

first

With

delicate

on something that would me.

I

didn't

It

seemed like

a

good one:

to other things. Ruoff

task

man-

a chance to

was

work

in Building 107,

lunchtime discussions with Wilcox.

me

but never vindictive. on, so

it

pass through the

what we were working

neers in Wilcox's group had their offices there.

active

would

su-

posi-

on,

about engineering philosophy.

The Robotics Laboratory was

putting

The

the Robotics group, and that Wilcox had taken over the

start conversations

ways

possible.

actually fly in space.

building once in a while, ask questions about

in

less likely to get

know either of them well. I knew that Ruoff had been the

when Ruoff went on

and

were

then one day Brian Wilcox and Carl Ruoff asked to meet with

pervisor for tion

work would be

encountered objects

to be the system engineer for the project. Well, that

new job I was offered.

And

it

that kind of feedback, things

much more

me

as

was

I

He had

learned that

a safe bet that if

it

I

and most of the engi-

sometimes found myself

a sarcastic wit that I

couldn't

occurred to

tell

me

was

al-

when he was that he

might

be joking, then he was. I

One

sat in the Building 107

of them

said,

"We

pen. Things get done."

room facing Wilcox and Ruoff. when you work on a job, things hap-

conference

notice that

They had

a task that

was languishing.

It

needed

SOJOURNER

40

someone

DDF

prod

to

"Blue Rover

it

into producing results.

JPL had

search and development.

virtually

The

funding, which the director practice, a

the

no moneys devoted

most

to

to self-directed re-

DDF was what little JPL had of this type of

was

free to allocate as

he or she saw

fit.

In

committee within JPL reviewed proposals submitted by em-

ployees, awarding

A

me

to

Compared

stood for Director's Discretionary Fund.

large companies,

ideas.

They described

DDR"

amounts of $10,000 or $20,000

couple of years before

had succeeded

most promising

my meeting with Wilcox and Ruoff,

in selling the first big

dollar, multi-year effort to

for the

DDF:

Carl

a several-hundred- thousand-

develop technologies for planetary rovers. The

principal selling point for the research

was

crucial to the future of the Laboratory:

that these technologies

Most of the

were

planets of the solar

system had already been explored by JPL spacecraft, or would soon be visited

by the Voyager mission currently in progress. Once the current phase

of deep space exploration was complete, the next step would be to ex-

moons in the solar system. Unless challenge when the time came, the Lab's fu-

plore the surfaces of the planets and

JPL was ready

to take

up the

ture funding, and indeed

its

reason to

exist,

could be in doubt.

DDF was far reaching in terms of the research questions to be addressed. How would a robotic arm mounted to a mobile platform pick up samples? How would the rover determine its location in the terrain using its stereo cameras? How would the rover recAs

originally funded, the

And how would it automatically plan a path around those the goal? The time was approaching when the DDF would

ognize hazards? obstacles to

need to demonstrate tangible answers to some of these questions. There were several research engineers participating on the DDF, mostly in our section, but also a few in Section 366, where Al-type

ware was supposed to be done

at JPL. It

was important to spread the work

around some within the organization. This was the biggest to this point,

and

it

gobbled up

a lot

soft-

DDF

ever

up

of the funds that could have gone to

other sections.

Ruoff and Wilcox were asking

me

to lead a

poorly funded, part-time one) that was signed up to

team

make

(albeit a small,

the Blue Rover

The Right Place

smart enough to drive

itself to

way

hazards along the

at the Right

destinations

Time

we

41

specified, while avoiding

instead of driving into them.

Once

became

this

The whole

task

was

just infused with a sense of turning science fiction into reality.

And

the

clear,

I

realized that

tools to

was

do

it

couldn't imagine a better job.

I

had already been mostly assembled.

All that

had

to be

done

to bring the pieces together.

DDF was "Semi-Autonomous Navigation." and yet— to my engineering soul—magical: Hidden

The crux of the Blue Rover

The name was inside

was

dull

a combination of technologies that

on its own

hicle to drive miles its

would enable

a robotic ve-

across an alien landscape without help

from

human masters.

SAN depended on

a repeated

sequence of

activities: sense-perceive-

Many of the functions in CARD that were done manually would instead be performed by sensors and software. The rover would now be

plan-act.

required to sense

its

surroundings, transform the raw sensor data into an

understanding of the safe areas and hazards in the

through the

and

terrain,

nearly to the limits of

over, effectively putting

its

own,

If

execute the plan.

had been able

one foot

many

The

rover

plan

its

path

would reach

to sense, then stop. For

SAN and

to repeat the entire process over

in front of the other, building

the rover could

could cover

it

it

would have

to be useful, the vehicle

yards of traverse.

finally

what

vicinity,

move only

a

up the

few yards per minute on

times the distance allowed by the simpler

CARD approach in the face of a forty-minute round-trip communications time delay between Earth and Mars.

Many of the orbiter that

maps of

mission concepts circulating at the time proposed a Mars

would take high-resolution pictures of the

surface,

producing

large areas distinguishing terrain variations of several feet. Since

the orbiter

would be

there anyway, rover operators

on Earth could

take

advantage of these maps to plan a "global" route for the vehicle, one that

extended

far

beyond what could be seen from the

eras. If the orbiter

eral miles

maps were

rover's

onboard cam-

extensive enough, perhaps rover routes sev-

long could be planned.

SOJOURNER

42

The map

for the terrain

for the route

through that

this point, the

surrounding the rover, together with the plan terrain,

would be uplinked

would be on its own. Just like

rover

capture a set of images with these pictures back to a

its

to the vehicle.

At

CARD rover, it would

a

stereo cameras. But instead of sending

human operator, onboard software would process map of the nearby terrain. "Machine vi-

the images to generate a "local"

would compare points in the left and right image views and

sion" software

determine the distances to features in the scene. These distances would then be transformed into an overhead map.

and

rover's view, there

available at

The

to the vantage point of

would be much detail in the portion of the map

the original images, there close to the rover,

Due

nearby rocks blocked the

less farther away. If

would be "shadows"

map where no

data

was

against the global

map

sent

in the

all.

local

map would now be matched

from Earth. "Terrain matching" the two maps was paper drawing of one

back and forth

map down on the

until they

were

aligned.

other,

like

placing a tracing-

and then

sliding the

Once you got them

maps

lined up,

you

map that was better than either of them alone: The local map gave you much higher resolution in the immediate vicinity, and the global map gave you coverage beyond the rover's sight. And once the maps were merged, the rover would know exactly where it was positioned relative to the global route plan it could merge the two of them together into a single

was

trying to follow.

The merged map would be analyzed for vation changed too

much from one map point to the next,

sign of either a steep slope or a big rock.

what

traversability If the

Without

really

this

map

was

a

ele-

good

understanding

a rock was, except as a collection of higher-elevation points sur-

rounded by lower-elevation ground, the software could divide up the map into hazardous

and

safe regions.

Here the automatic path planner would kick for paths

through the traversable parts of the

global route.

Once

it

found one,

it

The planner looked

map to waypoints

would pass

motion-control software, which, just as in the the rover to the waypoint.

in.

it

along the

along to the rover

CARD system, would drive

The Right Place at the Right Time

If

you could just do

this

43

over and over, without error, you'd be

home

free.

Navigating a physical vehicle in the real world forced the error.

team

to deal with

This was not really error in the sense of a mistake, but error in the

form of measurement Ideally,

wanted

it

you could

to go,

and

tell

it

however

destination,

uncertainty. a rover exactly

would

what path

to take to

where you

perfectly execute that path, arriving at the

distant, like a car pulling into a

parking space. But

even well-designed sensors have limitations in the precision with which they can take measurements.

Rover the direction

it

faced

have drifted two feet off

The counters on

feet.

its

the magnetic compass that told the Blue

If

was

off

by

just

one degree, the rover could

course by the time

it

the six rover wheels could measure wheel rotation

to within

two

rocks. All

wheels didn't cover the same distance:

rock,

had

had traveled one hundred

inches, but the wheels themselves could slip in sand or

and another rolled over

flat

a longer distance to cross to

rough, the rover might cover

on

one wheel went over

If

a

ground, the one negotiating the rock

keep up with the other.

more than

a

hundred

If

the terrain

was

feet to get to a spot

ninety feet distant. So odometry measurements were inherently uncertain.

The

result

was

that rovers never

went

exactly

where they were

sup-

posed to go.

The ally

deviation between a vehicle's planned path and the path

follows

Why

is

a

and

human

the car

is

don't the

same

street? If

errors apply

driver

is

when you

you closed your eyes

tried to get there blind, the

Wilcox knew

drive a car, or even just

after picking

your destina-

same problems would show

up.

it

to go.

dead reckoning error would be a problem, and had

planned to implement "visual tracking" software to compensate for the Blue Rover,

all

But

constantly correcting for the difference between where

going and where the driver intends for

straight-line

actu-

called "dead reckoning error."

walk down the tion,

it

it.

For

paths were defined as a combination of turn angles and

path segments the vehicle would execute. The visual tracker

SOJOURNER

44

would use images from the

rover's stereo

cameras to improve the rover's

execution of both turns and straight-line traverses. The tracker compared successive images. either

from

left

During turns, objects seen by the cameras would move

to right, or right to left

the direction of the turn. a

through the images, depending on

How far an object moved across the images was

good measure of how far the rover had turned. For straight-line forward motion, the tracker would compare "win-

dows" within successive camera images.

A

single full

cameras on the Blue Rover was composed of

TV

an array 320 pixels across by 240 pixels up and down, painting.

The

tracker focused only

on

image

dots, or "pixels,"

a small

like a pointillist

window, 30

came

in,

the original

by

20,

When

the

pixels

centered on the expected end point of the straight traverse. next image

for the

arranged in

window was checked against a window new image, and against new windows and down. Whichever window matched

in the corresponding region of the

shifted just slightly

left, right,

best with the original

up,

window

identified the direction in

and the rover could

had

drifted off course,

ing.

The best-matched window

window for

in the

which the rover

now improve

its

dead reckon-

new image became the

"original"

the next round of visual tracking.

Wilcox needed to upgrade the stereo camera mount he had originally constructed so tion induced

hastily.

when

The cameras needed

from the

to be isolated

the Blue Rover rolled over rocks. Otherwise,

vibraif

the

cameras jerked around too much, the visual tracker would lose lock and

become

useless for correcting the rover's dead reckoning error. Wilcox

designed a rection,

But he

new camera mount that allowed the cameras to pivot in any di-

and returned them to upright orientation using

still

had

to

the inherent limitations of the visual tracker. that

He came up

depended on wide paddles moving through

lasses.

The paddles would be

molasses-like stuff

a set of springs.

smooth out the camera motion enough to

fixed to the

would be mounted

stay within

with a scheme

a thick liquid, like

mo-

camera mount, and jars of the

to the rover body.

Wilcox stopped by the drugstore and bought several infant-sized baby bottles

and

liquid.

The baby

a can

of

STP

bottles

oil

treatment, which

would be

perfect!

was

He

a black, highly viscous

could

fill

them with STP

The Right Place

and use the nipples to got

some

seal the oil in so

odd looks

really

at the Right

at the

it

Time

45

didn't get

all

over everything.

checkout counter." Apparently the

checker wondered just what Wilcox was going to be feeding his baby

"The cashier

said,

gineers working

"I

That should fix the

sucker/ Really."

little

on the Blue Rover would

also tease

.

.

.

The other en-

Wilcox over

his un-

orthodox choice of off-the-shelf components. But the baby bottles did the job. as

"The camera mount stabilized the images just like

compared

motion of

to the jerky

a novice

a

good cameraman

new

with a

camcorder.

And

more than one

pixel per

of the Blue Rover upgrades were falling into place.

A micro-

the visual feature tracker didn't have to search

frame for the motion."

The

rest

computer had been mounted trol all

in the rover's

now image buffers installed to

of the onboard motors. There were

freeze images

from the rover cameras,

middle compartment to con-

as a true

CARD

system required.

We

still

live

video from the cameras had to be transferred over a long cable back

could not afford to

install

the buffers

on the

vehicle

itself,

so the

to the control station.

To

test the

from the

new driving algorithms in the lab,

ceiling

by braided

steel cables.

the rover

was suspended

S-hooks on the ends of the cables

slipped through rings bolted to either side of each compartment, six in

The

rover

would hang

anywhere.

When

straints, drive its

it

in the

air,

all.

wheels turning, driving but not getting

attempted a turn, the vehicle would twist in

its re-

forward and rear wheels, then twist back. The suspended

rover was vaguely reminiscent of a cross between a puppet

some kind of medieval

on a string and

torture device.

# "Hard

right.

Hard right

walkie-talkie. Brian

.

.

.

No! Hard

Cooper

left!"

rolled his eyes.

ing Wilcox's jerry-rigged control station the Blue Rover in response to Wilcox's

Cooper swung the joystick

Wilcox's voice

over,

—

He was

the

the engineer upgrad-

but right

commands

came over

now he was

driving

radioed from the

field.

and waited to hear the next instructions

over the walkie-talkie. After a few seconds, the images on the monitor in front of

him panned

wildly "Stop! Stop!" Cooper complied, hitting the

SOJOURNER

46

red emergency button.

He shook his head.

Brian Wilcox was a brilliant en-

gineer and a good supervisor, but he was lousy at directing the move-

ments of the Blue Rover to

They were

CARD tion in steel

trying to

move

the rover into

its

starting position for

Arroyo Seco. Getting the rover there was

testing in the itself.

remote operator.

a

The Blue Rover would start out suspended in the

a produc-

air

from

six

wires for in-laboratory testing. First a couple of engineers would

drop the

rover,

one wheel

at a time,

from the wires onto

wooden pallet. Then they'd push the pallet over to the south face of Building 107,

door had

lifted

hand operate the

wheeled

a

the big garage door

on

roll-up chain until the

high enough for the rover to pass, and push the pallet out-

side into the alley beyond.

The

rover

was then carted down the

onto Surveyor Road, stopping traffic, rolled past the guard the JPL East Gate, and lifted

station,

up onto the sidewalk on the west

it

was

safely

through

side of the

road-bridge that crossed the Arroyo Seco into the east parking rover didn't cross the bridge. Instead, once

and

alley

lot.

The

out of the road-

way, the engineers shepherding the rover plugged in the fifteen-hundredfoot umbilical cable that

would allow commands and video

between the Blue Rover and the rover control

to pass

station in Building 107.

From here, Brian Cooper could now teleoperate the rover at its stately pace down the grassy slope into the arroyo, taking direction from the engineers in the field to ensure that he wasn't putting the vehicle at

Brian Wilcox had

first

risk.

driven the Blue Rover decades before, and un-

derstood intimately the issues of time delay that arose

when a human op-

erator tried to control a machine thousands or millions of miles distant.

But even when the rover was

hundred yards away

in the arroyo

in the laboratory, the

the lags in verbally relayed instructions as if the Blue field,

a

few

slow steering of the rover and

made

for vehicle response delays

Rover were on the Moon. But since Wilcox was out in the

standing a few feet from the rover, the reality of those delays just

wasn't registering.

He was just

he was expecting

to,

it

seeing that the rover wasn't going where

so he had to continuously give

Brian Cooper. Cooper had learned by

now

Wilcox's directions, the rover's motions ratic

and the driver was only

new instructions to

that if he blindly followed

would

get

more and more

—just as Wilcox himself had observed long ago during his

first

er-

expe-

The Right Place

rience driving the

SLRV The view from the

ited to a thirty-degree swath, not

Cooper

so

it

up

enough

He

to teleoperate the vehicle.

right to build

at the Right

wanted

4?

Blue Rover's camera was lim-

much

to provide

a ninety-degree field of

view but

that

left

would take too

and

long,

from someone out in the field.

do was get the vehicle into position so the

to

context for

could always steer the rover

could be really helpful to get instructions

All they

Time

real testing

could begin. Wilcox was about to direct Cooper to joystick the rover closer to the starting line for the first.

CARD test.

"Can you put Steve on?

mann was one Steve got

I

need

Over the to ask

walkie-talkie.

him something." Steven

of the other engineers participating in the

on the

line,

Cooper was

field test.

Katz-

Once

"Whatever you do, don t

conspiratorial.

give the walkie-talkie back to Wilcox.

Cooper spoke

Now tell me which way to drive."

By the

summer of

CARD

system. Brian Cooper could designate a target forty yards in front

1986, the Blue

Rover team had demonstrated

a real

of the vehicle, and the rover could hit that target within about two yards.

That wasn't too bad, and

their analysis

showed

cameras they could do better or designate

that with higher-resolution

farther.

The

visual tracking sys-

tem, even with the aid of STP-filled baby bottles, did not perform so well. It

ran too slowly on the computer then available, and was abandoned.

Proving the

SAN concept took longer. The

thirty-degree field of view t

enough informa-

machine vision software

for the automatic

of the rover cameras was just too narrow. There wasn tion in the output of the stereo

match. So Wilcox took the baby-bottle camera

terrain

matcher to get

mount

off the rover, installed a pan-tilt head,

eras onto the head.

By turning

a

and reattached the cam-

Now the rover control computer could aim the cameras.

the pan-tilt head

left,

right,

and

straight ahead,

three sets of stereo pairs, the rover could collect a

and capturing

panorama with

a nearly

ninety-degree view. Just as

we were

getting ready for field

tests,

our research partner from

Division 36 reported that his automatic path planner didn't work.

developed tions.

a clever

approach to storing the terrain

But the path planning algorithm he had

had turned out

to

be

a

dead end.

map

at

designed to

He had

multiple resolu-

run on

this data

SOJOURNER

48

Don

Gennery, a Ph.D. from Stanford and the engineer in Wilcox's

group behind both the stereo vision software and the

terrain matcher,

had

seen journal articles about various planners, and didn't think that path

planning should be so

difficult.

He

certainly didn't

of what he viewed as a minor element of

own

ing the veracity of his

wrote the software code

SAN

want

to see the failure

stand in the

The SAN

for a working,

mathematical

was slow to run. But

test

though computationally intensive

it

all

of Gennery's

worked.

runs were excruciatingly slow. Gennery believed in

rigor.

He was

the group's algorithm master. But he had

interest in optimizing his software for speed.

view, proving the concept

plementation

prov-

algorithms and software. So in two weeks he

planner based on an algorithm he had read about. Like code, the planner

way of

was

From Gennery's

all-important; everything else

no

point of

was just im-

So the machine vision and terrain matching pro-

details.

cessing took hours. If

we were

lucky,

we

could get maybe three

SAN

day out in the hot dusty arroyo. Members of the team

cycles in a full

spelled each other in standing guard over the Blue Rover. In the end, after a lot of literal sweat, the Blue Rover proved the

SAN

number of sense-perceive-plan-act cycles, and crossof ground. And a JPL video crew got it on videotape.

concept, performing a

ing about

fifty feet

We had made the Blue Rover navigate in the Arroyo Seco.

Rovers were

small change. Every few months,

was ready to work on one of the

ask

if

too

late.

I

still

I

came up

my

supervisor

would

"big" robotics projects. But

it

was

had already caught the rover bug. So each time the opportunity to

work on

a robot that rolled rather than reached, or drove to

new places instead of being bolted to the floor, I took it. More money arrived from TACOM, and a new "rover" came with it.

The

vehicle

was

a military

Humvee,

the replacement for the venera-

ble jeep.

Out of the

success of the

CARD demonstration, TACOM had funded

a newer, larger task at JPL, the Robotic

The new

RTTV

Technology Test Vehicle program.

team, comprised mostly of engineers

on the Blue Rover, was

to

reimplement

CARD

who had worked

on the Humvee,

simul-

The Right Place

at

the Right Time

49

taneously creating a system that could evolve to greater and greater capability.

TACOM also wanted the new system to be demonstrated around the country

at

Army

The

into a semitrailer. to protect the

new

mobility test courses. So the

was carpeted and air-conditioned

trailer

computer workstations

that

(primarily

were located inside).

the comfortable yet windowless de facto offices for Brian

Steven Katzmann.

went

control station

It

became

Cooper and

When needed, the trailer could be hitched to a tractor-

truck and hauled anywhere.

Humvee

Switching from the Blue Rover to the challenges.

The Blue Rover drove very slowly.

created a few

paths consisted of a

Its

new

num-

ber of straight lines strung together. Whenever the rover changed course, it

would come

to a stop, steer

drive

through the turn to

tinue

on along

a

new

its

three-compartment body into a turn,

its

new

heading, stop, straighten out, and con-

straight line.

The Humvee needed

to

move

faster.

so the

CARD paths would now

have to be curved, and no more sharply than the

Humvee could negotiate

No vehicle in motion could turn instantly, at speed.

The cial

CARD control station went through a transformation. Commernow catching up with our needs. We junked the dim

technology was

3-D display that

relied

on polarizing

filters,

carefully

fied television monitors, half-silvered mirrors,

Silicon Graphics Inc.

station that could

mounted and modi-

and polarized sunglasses.

now manufactured a color graphics computer work-

become

a stereo vision display

The computer

screen

could alternate back and forth between displays of two different images,

and

special battery-powered goggles

low only one eye

with liquid-crystal shutters would

to see the screen at a time.

The goggles were

al-

electroni-

cally synchronized to the display, shuttering the right eye at the exact

instant that the versa.

image meant

Even though

at exactly the

twice as

a

for the left eye

was being shown, and

person wearing the goggles never saw both images

same time, the images were updated or "refreshed" so

fast as a

fortable to look

normal

at,

vice

and

TV picture —that the

display

clearly three-dimensional.

The operators didn't get

headaches from watching the screen, which had often been with the old system. Alignment of the

left-eye

fast

was smooth, com-

a

problem

and right-eye images was

SOJOURNER

50

no longer an

because both were being shown on a single monitor.

issue,

There was no way

them

for

was the best 3-D

total effect

was

Brian Cooper's job

new

to get out of alignment with each other. display any of us

The

had ever seen.

new CARD control station on The SGI machine was designed

to build the

Silicon Graphics workstation.

the for

graphics, so instead of crude crosshairs to designate waypoints, this time

Cooper would

create a

turn the graphical

3-D model of the Humvee.

He would be

able to

Humvee in any direction, and place it anywhere on the

3-D terrain display he wanted to. As he pushed on the joystick to move the

model ever

farther out into the terrain, the computer-generated

would get smaller and

smaller, just like the real thing.

be able to use the model to

two

obstacles, or

tell

would have

whether the

Humvee

You would

actually

would fit between

real vehicle

go around them. And instead of merely

to

marking the positions of waypoints in the images, the control station would display the entire path as a

winding road (with yellow

bricks!) shrinking

into the distance.

Cooper had been an

officer in the U.S. Air Force.

paid for his college degree and promised to put

good

He had

use.

that he

was stuck managing tasks

—

for

which he had no

When

tunity to

become an

new SGI

machine.

It

was

this

as if

What more

of the employees

new hires some

Among

new to

a

new 3-D

kid's toy at the other end,

could he ask for?

Every year, the Electronics and Control division

division.

Now he had the oppor-

he was getting to design

one had the ultimate

camouflage -painted Humvee.

the

—instead

expert at the software tools available for the brand-

computer game, only

all

interest

he came to JPL, he soon found

himself upgrading the Blue Rover control station.

briefing for

Air Force had

gotten out as soon as he could once he discovered

of doing real engineering work.

a

The

his engineering skills to

would conduct

the division.

The

idea

a daylong

was

to give

background on the organization and goings-on of the

the several presentations

was

a talk, usually given

by

Brian Wilcox, on "Autonomous Vehicle Research at JPL," which detailed the told

CARD me,

and

SAN

technologies

we were

"Why don't you do it this time?"

developing.

One

year Wilcox

The Right Place at the Right Time

The conference room was

filled

with

my

turn to speak,

was

referring to a chart that consisted

fifty

51

or sixty people. As

I

waited

watched the presentation before mine. The speaker

I

of columns of numbers. The num-

—they indicated how much money was coming into —but knew wouldn't be able to various types of

bers were important

JPL

for

member I

activities

anything from the chart once

didn't have

was

disappeared.

any charts with numbers.

My turn came. It

it

re-

I

I

I

placed

my first viewgraph on the overhead projector.

a color picture of the surface of Mars, taken

by the Viking

1

lander.

Reddish-hued rocks were strewn over the landscape, stretching from the foreground into the cles at

JPL because

far distance.

we want

over that horizon ..."

to

"We do

research

on autonomous

vehi-

go here. And we want to find out what's

FIVE

THE BIG ROVER THAT NEVER WOULD

the last few years of the 1980s, rover research began to heat up.

Inwas funded to

do

a

new

study for what might be

its

next big mission:

Mars Rover Sample Return, or MRSR. At about the same time, also started

up Pathfinder Planetary Rover,

gram designed to develop

a

the technologies that

new

JPL

NASA

rover research pro-

MRSR just might need to

succeed.

The centerpiece of the JPL research program was to be Planetary Rover Navigation Testbed vehicle. As

new

the Pathfinder

technologies ma-

tured, they

would be migrated

for tryouts.

The new rover was going to be big, much bigger than the Blue

Rover.

The

off of the

increased dimensions of the

bench top and onto the testbed

new

were not driven by

vehicle

any mere showmanship desire for impressive physical

size,

but by a piece

of equipment called the nineteen-inch rack.

The purpose of

the rover

was

to be a testbed for

improving and

uating navigation approaches like Semi-Autonomous Navigation,

and whatever

else

came

next.

a

CARD,

What the Blue Rover had lacked as a testbed

was onboard processing capability: The end of

eval-

rover's "smarts"

were

at the

other

long cable leading back to Building 107. Future rovers operating

on Mars would have

to carry their

own computing.

computer power necessary to operate

Proving that

a rover could

all

of the

be carried by that

The Big Rover That Never Would

rover

would be

the vehicle

a

53

key feature of the research program. By the same

would

also carry

its

own power

logic,

source onboard, capable of

powering both the computer and the motors.

The

testbed chassis

Like the earlier vehicle,

would be it

loosely based

would have

six

on the Blue Rover

wheels and three compartments.

But one problem with the Blue Rover stemmed from the steel that

rain

connected the three bodies. Keeping the rover

might require knowing the

design.

relative orientations

flexible spring-

safe in

rough

ter-

of the compartments,

so the vehicle could stop before one of the compartments could tip over.

There was no easy way to mount a sensor on the the

spring-steel

new rover would replace it with rotational bearings

coders" that could precisely measure the

What would be

the

most

tilts

member,

so

outfitted with "en-

of the compartments.

cost-effective design for a navigation test-

bed? As Wilcox reasoned, the vehicle was primarily a platform for moving the navigation sensors and computing subsystem around.

would be capable of

driving into

The

rover

somewhat rough terrain, but only to

the

extent necessary to fully exercise the navigation software. Since computers

and related electronics were improving year by

that the

year,

it

seemed

likely

onboard computer Central Processing Unit would be replaced,

perhaps several times, before the mechanical chassis was considered obsolete.

The

boards and

testbed had to allow for easy change-out of image processing

new CPUs,

as well as other electronic

The most

eventually be selected.

readily available standard for electronic

equipment mounting was the nineteen-inch card cages and for

power

rack.

Once

room rest

equipment would

could easily find

racks.

Two racks would

for expansion.

the decision to place

been made, the

One

supplies with holes pre-drilled into their faceplates

immediate mounting into these standard- width

leave plenty of

equipment that might

two nineteen-inch racks on the rover had

of the vehicle began to

certainly overheat during

summer temperatures

fall

together.

outdoor

The

electronic

testing in the

90°F+

of Southern California; an air-conditioning unit

would be mounted on the back of the

racks.

The only

low-cost portable

source of power that could provide the kilowatts needed for the electronics

and air-conditioning was

motor generator, which could be Like the Blue Rover, the testbed would

a gasoline

mounted elsewhere on the vehicle.

SOJOURNER

54

also require a set of stereo

cameras to sense the nearby

terrain, preferably

with an unimpeded view of the surroundings. The electronics rack was already the tallest point

on the

would go on top of the

rack.

want

tually

gated to a

would be

arm

to

site,

vehicle, so a pan-tilt

And because

head and camera bar

would

even-

do something useful when the vehicle successfully

navi-

the research task

such as pick up a rock sample, an available robotic arm

affixed to the front.

to act as the rover's

You

didn't really

bumper, sticking out

want an expensive robot

in front as the first point of

arm was

contact between the testbed and a big rock. So the mechanical

mounted on

a

ward when

was needed

it

"dunking bird" assembly that would rotate the arm for sampling,

way during traverses. The testbed design matured. When over twelve feet from end to end. tires thirty-five

work

it

was

it

built,

would need

back and out of the

the testbed

would be

six recreation vehicle

inches in diameter to support the vehicle.

We

went

to

requisitioning components, having the chassis elements machined,

assembling the pieces. tle

We

and rotate

for-

We

had

a contest to

name

the testbed, with a bot-

of champagne for the winner. Despite the plethora of names submit-

ted,

we ended up with

since the

"Robby," seemingly the moniker for every robot

movie Forbidden

Planet.

Although Robby's size had been based

on

who saw

vehicle

and one that would actually be sent to another

common

the rover

seemed

many

practical requirements,

people

to miss the distinction

questions were "Will those rubber

tires

between

planet.

a test

The most

work on Mars?" and

"Do you really expect to send something that big?" We even contemplated spray-painting the tires silver to circumvent some of the inquiries. The reactions to Robby caught Wilcox off guard. "Many people who saw the Blue Rover So

I

said 'Oh,

was very surprised

it's

so small.

I

envisioned something bigger/

at people's reactions to

exactly twice as big. Their reaction

was always

Robby, which was almost

'It's

so big.' There

must be

a narrow band in between which would have generated no reaction

•

at all."

The Big Rover That Never Would

of the

All

work on Robby was geared

stone." JPL

had signed up

to prove that

to the

55

"One-Hundred-Meter Mile-

Robby could autonomously traverse

the length of a football field through the natural terrain of the Arroyo Seco.

Once

group partnered with the AI-

again, the Robotic Vehicles

focused Robotic Intelligence group to create the software that would

make Robby go. Wilcox s team would develop the sensing and perception software that would build terrain maps from the raw stereo images that would come out of Robby s cameras. They would also do the terrain matching and the actual control of Robby's motors. The Robotic Intelligence group, led by David P. Miller, would write the path planner.

Donna at

JPL

Shirley led the

MRSR study team.

to have ascended through the

neers to a position of high the

sixties,

working on

visibility.

She was one of the few

women

male-dominated ranks of the engi-

She had been

several flight projects,

at the

Laboratory since

from Mariner

to Cassini.

Now she was running the effort that was the first nascent step toward the start

of a

new mission.

The study team included engineers from throughout

the JPL organi-

zation: mechanical, power, thermal, telecommunications, electronics,

and

computing engineers; interplanetary trajectory designers; mission operations people

—and

rover navigation and control engineers.

team met weekly. At pervisor with too

first

The study

Brian Wilcox attended, but he was a group su-

many other responsibilities, so the assignment of reprefell to me. Don Bickler was my counterpart for

senting vehicle navigation

rover mobility design. Shirley clearly enjoyed being in charge, the only

of male engineers. treated

them

as a

Much

woman in a room full

of the team was young, and Shirley sometimes

group of children

in her charge, requiring her guidance;

they were intelligent but naive, creative but needing seasoning.

The

objective of the

Mars Rover Sample Return mission was

ten pounds of Martian rock and

would be

soil to

to collect those samples

the Earth.

The

from geologically

to return

rover's job in this

diverse sites

on Mars,

then bring them to a lander where a rocket waited that would launch the treasure into

Mars

orbit

and then onward to Earth.

SOJOURNER

56

Shirley started the

MRSR team out by forcing the

engineers into do-

ing "trade studies." She was trying to hold the team back from rushing to a single design too soon. idea,

Often a creative engineer would

hit

on

a clever

then converge on a point design that represented an elegant solution

to the

problem he or she had been presented with. Unfortunately, the best

design for an isolated widget might no longer be best in the context of the entire system.

made the team consider a range of options for each of the key technologies. The MRSR rover might use "structured light," laser range Shirley

finders, sonar, or stereo vision to see

what hazards confronted

it.

How

many wheels should the rover have? Six seemed like a good number, based on the Blue Rover and Bickler's work. But why not two, four, or eight? Maybe the best rover would have no wheels. After all, humans had a lot more personal experience walking on legs than rolling on wheels. And

much of Earth's terrain was off-limits to wheeled vehicles, on

The study team members

foot.

reachable only

representing each rover subsystem ex-

plored the range of design options available in their areas of specialty.

The plan progressed: drive for

for

The

MRSR was ambitious, was

rover

and became more so

as the study

to be a rolling geologist's laboratory.

It

would

hundreds of miles across the surface of Mars, then be directed by

scientists

on Earth

up

to pick

samples for return to Earth.

rocks, drill cores, slice, dice,

and package

A mapping orbiter flying overhead would cre-

map for the rover at three-foot resolution; using that map, the rover would always know exactly where it was on Mars. It would be so smart that it could drive for miles without guidance from human operators, and might even choose to call home only if it figured out that someate a terrain

thing

it

saw was

scientifically interesting.

artificial intelligence,

would be

MRSR would push the limits of

computing power, and robotic navigation. And

heavy, hundreds of pounds.

Once

it

had collected

all

that

it it

could hold, the rover would select which samples to keep, and which were

no longer

exciting

enough

to hold

on

to.

The

rover

would

find

its

way

to

the sample-return lander, and transfer over a sample canister with several

pounds of rock and back to Earth with

its

soil. Finally,

the return vehicle

processed alien cargo.

would launch

itself

The Big Rover That Never Would

5?

m Robby was not to be the only focus of the Pathfinder Planetary Rover pro-

gram.

Don Bickler finally got the money to build mobility models. He and a team of mechanical engineers wanted

to prove that the mobil-

performance of the small models would be

"scale-invariant." If the

few of ity

his

rock-climbing and stability characteristics of tabletop rovers could be rectly extrapolated to their full-sized versions,

then the mechanical engi-

neers would have a relatively cheap and easy approach to trying out mobility concepts.

seemed

It

tion,

they wanted to prove

ler's

guys built Robby

plausible,

in the

summer of

1989 Bick-

They

a one-seventh scale version of Robby.

Jr.,

new

but before they accepted the no-

by example. So

it

di-

would come into work in the morning, then spend the afternoons in Bickgarage machining the rover components.

ler's

from

Eisen, a co-op student

Eisen thesis.

made

He

rails

formulated mobility

He

delivered to 107.

slopes of various scale-invariance,

drives

M.I.T.

the study of the

Robby Jr., then with more

tilts

Among them was Howard

two Robby s the tests,

difficulty

set

subject of his Master's

then conducted them easily on

on Robby. Eisen had huge aluminum

them up

at several different angles, creating

up which Robby would

drive.

The work proved

and incidentally showed that Robby's

were too weak

for

all

existing

wheel

but modest traverses in outdoor terrain. Up-

grading the torque capacity of the wheel drives to extend the testbed's operating environment

became

a

new

milestone. Eisen

went back

to M.I.T.

to complete his thesis.

Since building the pantograph rover model, Bickler had continued to

toy with vehicle concepts.

graph a

He wasn't satisfied with the trouble the pantocould get into when attempting to drive over

—or the Blue Rover—

"bump"

rather than a "step." Bickler wasn't crazy about computers or

software, and liked to

rile

up Wilcox and

his

group by saying

that a

mobility system wouldn't need any of "that onboard intelligence

good

stuff."

But sometimes computers did have their uses. Bickler found himself

tin-

kering with designs by computer analysis, rather than building every

SOJOURNER

58

minute variation he came up with. One of the tried

six- wheeled

was much simpler than the pantograph, with

end of

a larger

master bogie or "rocker."

designs he

a smaller bogie at

one

He played with the proportions

of the bogies and the sizes of the wheels. The vehicle's hazard-crossing capability ler hit

seemed highly dependent on these proportions. Eventually

upon

a version that the

computer

analysis said

Bick-

would outperform

the pantograph. In

November 1989

Bickler's garage

produced another rover model,

with Bickler's crew of young engineers providing Imagining Robby to be a

Mars

full-sized

much

of the labor.

rover, they sized the "rocker-

bogie" to be a one-eighth scale model. The rocker-bogie rover worked as the computer analysis had predicted, climbing as well as the pantograph

over steps, and surpassing the pantograph pler

and better! They

really

show it

took an

interest.

off.

model rover over

to

Donna

Shirley's

When he came into the front area, Shirley's secretary

She thought

ler called the rover,

Sim-

had something now!

Bickler brought the motorized office to

when driving over bumps.

it

was

She wanted to know what BickHe didn't really have a name for it,

cute!

and how it worked.

so he started to explain the running gear: "You've got the rocker here, see,

and then

Bickler

this front

bogie

.

.

."

"Rocker-bogie" became "Rocky."

was proud of the rocker-bogie

design, but he worried that

it

might

not be good enough.

NASA had directed JPL to fund other organizations

in addition to itself

through the Pathfinder Planetary Rover program.

NASA

Congressional mandate required that a large fraction of

research

funds be directed to private industry and universities, and the headquarters

manager responsible

money should be

spent.

for Pathfinder

had

specific ideas

The consequence was

was funneled through JPL

that

about where

his

much more money

to other organizations than stayed at the Labo-

ratory.

The primary

beneficiary of these research dollars

Mellon University's Field Robotics Center thought about

it,

that tasted rotten.

his expression

was

as if

in Pittsburgh.

was Carnegie-

When

Bickler

he had bitten into something

"The big money always went

to

CMU,"

said Bickler.

The Big Rover That Never Would

"JPL got a hundred thousand, and

was

research

was the research community's Pittsburgh team's forte

Whittaker was

On

that end.

day,

principal

CMU's

proponent of "big"

was teleoperated dump

robotics.

The

trucks, earthmovers,

and

hazardous environments.

He wanted nothing less than the ubiquitous use human society, and would do anything to achieve

first

meeting him,

and sleep only

I

imagined that he must work twenty

a couple of

hours a night, an incongruous in balance

by force of

Whittaker seemed to possess an army of "slave" laborers students and recently degreed engineers

whatever concept Red proposed. For

walking robot with

"You don't

six sixteen-foot

really think you're

Over the prior twenty eral institutions,

years,

will.

—graduate

—ready to design and construct

NASA they were building Ambler,

a

telescoping legs. Bickler hated the de-

going to send

Ambler did make an imposing

the advancing

who

a zealot.

combination of exhaustion and energy, kept

sign.

robotics

hulking ex-Marine

a

for high-radiation or otherwise

of robots throughout

hours a

a million!"

by William "Red" Whittaker,

led

rugged vehicles

CMU got

59

this

thing to Mars!" But

display.

walking vehicles had been created

at sev-

and some of them were impressive, going places wheeled

vehicles could not. Mechanical complexity

and low power

efficiency

the bane of walkers. Electromechanical legs were inherently

were

more com-

plex than wheels, usually requiring several motors to the wheel's one.

Walkers had to maintain their balance, and figure out where to place their feet in

lower

rough

its

terrain.

And because

body with every

which meant

it

it

was

as if

a rolling vehicle.

this assertion. It

finally

was only

admitted the

when Ambler

would indeed always be climbing

shifted

uphill,

Ambler faced the same weight problem rying

man

all

of

its

builders.

weighed over

six

CMU's Whitconstant

at a

never did carry

its

after

many months

reality that

its

of

each leg was

weight to that

leg;

Am-

burning energy.

as

Robby only more

computing and sensors onboard, It

stairs,

"glide" over flat terrain efficiently. Bickler ar-

development that Whittaker sinking into the ground

and

raise

were always climbing

it

Ambler design kept the main body

Ambler would

gued vehemently against

bler

walking robot tended to

was working harder than

taker claimed that the height, so

step,

a

it

stood far above

own power

thousand pounds, more than

a

so.

supply,

Car-

its

hu-

and even so

ton heavier than Robby.

SOJOURNER

60

One the

at a time, its legs

would

and swing around, then

lift

set

down

again,

body ever so slowly moving forward. Bickler feared that JPL

came

to build a real

Mars

was going

to lose

its

CMU would get the

rover,

when the time prize. He wouldn't

edge, that

mind so much if victory went to the institution with the best technical design,

but

it

seemed

to Bickler that

"CMU was getting all the money to do

stupid things!"

Robby and Ambler would

eventually meet.

They would

face each

other across a simulated Martian terrain during the NASA-sponsored

Rover Expo

in early

September 1992, an event showcasing robotic rovers

from around the world. The Rover Expo would be staged on the mall Washington, D.C., across the

street

in

from the National Air and Space Mu-

seum. Robby and Ambler would be the giants of the show. By then, however,

both would represent the

past,

not the future.

Despite the superiority of Robby's hardware over the Blue Rover, Semi-

Autonomous Navigation remained a slow, disturbingly time-intensive process. The bugs were being fixed. After integration and testing, Robby's hardware and software were

now functioning as designed.

Yet there were

not enough daylight hours to traverse the required hundred meters (about a hundred and ten yards) in a single day.

manage was

Don

The

best

Robby could

thirteen yards in four hours of continuous operation!

Gennery's software was the

culprit, requiring

over an hour for

stereo processing and terrain matching. Brian Wilcox tried to convince

him to put effort into reducing the software processing time, but Gennery would have no part of it. "The first version of Robby was all Don's code. I

couldn't light a

fire

under

Don to speed up the code and to

optimize

it

in

And it was causing us we didn't know how to run projects, and everything else, which anguish." Gennery was an extremely caused Don and considerable any way.

all

kinds of political problems. People were

saying

I

.

.

.

able engineer, capable of producing

major

results in short periods of

time. But he also suffered from "Ph.D. arrogance," devoting himself only to those activities that

he himself deemed worthy of

his talents.

Gennery

The Big Rover That Never Would

61

thought any reasonably intelligent person could recognize from the ing

Robby

key principles of

capability that the

extrapolate the performance of ers that

would obviously be

mized code

exist-

SAN had been proven,

Mars rovers assuming the

faster

available in future years, together

and

comput-

with

opti-

that other software engineers could generate.

Wilcox was

regularly having to explain Robby's rubber

still

couldn't expect people to do

important. In the minds of too

You

much extrapolating. Appearance was allmany people, slow performance translated

to "It doesn't work," regardless of the demonstration

We would have to do the demo again, Over the next

tires.

and get

s

technical merit.

right this time.

it

months, Robby's thought processes went

several

through a readjustment. Larry Matthies, a Ph.D. in computer science

from CMU, had recently been hired into Wilcox's group. He brought with

him

his

own

stereo vision algorithms and software,

which he adapted

for

Robby. Unlike Gennery Matthies was interested in speeding up his code.

The

terrain

matcher module was dropped altogether.

When Robby went than before.

It

out into the

field again,

it

was many times

navigated the One-Hundred-Meter Milestone in a

faster little

over four hours.

~w Robby was the only game in town. Dave Miller wanted to change

At J PL, that.

He had his own views

about

how rovers should work, views that did

not jibe with Wilcox's sense-perceive-plan-act paradigm. But

if

he was to

wrest control of rover research from the electronics and control division, Miller

would have

tute of

to start out small. Literally.

Technology

Artificial Intelligence

At the Massachusetts

Laboratory,

Rodney Brooks was

experimenting with tiny robots. Rather than try to duplicate ligence in any form, Brooks

much more modest yet insect tures

on

life

was attempting

creatures. Insect brains

swarmed over

the planet,

to

had

among

Insti-

human intel-

mimic the behavior of relatively

the

most

few neurons,

successful crea-

Earth.

Brooks's concept was that one could layer a ple behaviors,

number of extremely sim-

one atop another, and produce useful

results. Miller didn't

SOJOURNER

62

want size

to shrink rovers

would

down to

do. Miller's

Brooks's realm of robotic insects. Shoebox

group

built a four-wheeled tabletop rover they

named

"Tooth."

strate a

few behaviors, responding to

would never

It

survive outdoors, but signals

it

could demon-

from simple photo-detectors

and contact switches. Miller

was desperate

for a terrain-capable rover.

mechanical team, there were

now

Thanks

to Bickler's

ready-made vehicle suspensions just

waiting to be used. Miller and John Loch, an engineer in his group,

added

computer

a simple

Rocky

to the

vehicle to control

its

wheel

first

drives

and steering motors. They could then joystick Rocky around, driving over rocks and impressing onlookers. Bickler's experimentation

called built.

"Rocky

Rocky

But that design had been scrapped before

2."

3

with Rocky had led to an improved design

had come

wouldn't sink so easily in

it

was ever

off the assembly line with bigger wheels that soft sand,

and proportions optimized

for climb-

ing over rocks. Other people might look at Rocky 3 as a model of something bigger; to Miller, the vehicle was already full-scale, a "microrover"

ready to compete with Robby. Miller

needed to convince the

right people of the flaws in the

Robby

"big rover" approach, and of the advantages of the microrover. So he lob-

bied the JPL manager of Pathfinder Planetary Rover.

He

lobbied the

NASA headquarters sponsor of the activity, and he lobbied Donna Shirley. Miller carefully

promoted the revolutionary nature of the microrover:

would be cheap, what he

cute, sexy,

and imbued with AI magic,

called "Behavior Control."

Megarovers

like

in the

It

form of

Robby- were slow,

lumbering, and overcomplicated. Ironically, the

work of Rodney Brooks

idea of microrovers

had

itself

been

NATO

robotics

workshop

had inspired Miller

partially triggered

few years before. Wilcox had presented at a

that

to the

by Brian Wilcox

a

a "Micro-Lunar- Rover Challenge"

in Portugal in

May

1987, several

months

before Miller joined JPL. Wilcox had proposed the idea of sending rovers

massing

Mars

less

rovers

than twenty-five pounds to the Moon,

were expected

taken the ideas

JPL

home from

to support the

ongoing

to

at a

time

when future

be over two thousand pounds. Brooks had

the conference, while Wilcox had returned to

MRSR rover studies.

The Big Rover That Never Would

At the direction

63

NASA sponsor of Pathfinder Planetary Rover,

of the

ond prong was added

to the research program's focus:

a sec-

Demonstrate

a

microrover that returns a sample to a simulated lander. Miller's Robotic

group already had

Intelligence

Rocky

ler's

3 vehicle. Rajiv

computer on wheels, using

Desai built a crude

team mounted on the

that the

a simple

hardware and software into

a

front of

Rocky

arm with

3.

a

Bick-

sampling scoop

Their team integrated the

new rover system.

When all was ready, the rover autonomously dead reckoned to a specified target, its

using a compass and measuring wheel revolutions to estimate

Once

position.

the onboard computer had estimated that the rover had

reached the sampling the

arm onto

rather than

site, it

the surface.

halted the traverse and dropped the end of

To ensure

that the gripper

on a rock (which would probably not be

had

fallen

on

soft soil

retrievable) the

scoop

contained a pin sensor. The pin was designed to activate a micro-switch. the pin struck a hard surface, like a rock, the switch

and the rover would spot.

instead, the pin

If,

and the

tivated,

was

raised,

its

site

the

arm and drop

had entered

would be

at a possibly

it

soft soil, the switch

When it

more

triggered,

favorable

would not be

would be deemed "good." The scoop

and the rover traversed back to the

frared beacon.

of

lift

If

closed, the

ac-

arm

homing in on an indumped the contents

lander,

got close enough, the rover

scoop into a cardboard box that was a stand-in for

a lander's

sample

collection bin.

The

MRSR mission study had become

to justify

its

latest research efforts,

creased funding. tors for

grown lect

up

means

for

and market

each research discipline

its

NASA

sponsors for

in-

rover, carrying radioisotope thermoelectric genera-

power, multiple robot arms, and a sample-processing system, had

to 1,100 pounds. to ten

The lion.

The

a

An

It

would cover great

pounds of rock samples

price for those rocks

outside contractor

and

end

col-

was $10

bil-

in the

for return to Earth.

would be

was

distances,

high:

Our

best guess

hired to produce an independent cost es-

timate of the JPL concept. Their

numbers were worse, anywhere from

SOJOURNER

64

$10 to $13

billion.

That was

big, five times bigger

space mission we'd ever done.

And the

support for

than any robotic deep

MRSR started to evap-

orate.

Too

late,

we

realized the day of the big mission

the dinosaur that could not adapt to the teroid

punched through the atmosphere

colder, the skies

was

over.

new environment into the ocean.

MRSR was

after that as-

The weather got

clouded over, and the advantage went to the small.

SIX

THE LITTLE ROVER THAT COULD

Rover Sample Return had collapsed under

Mars This

left a

huge gap

in

NASA's plans

own

its

weight.

for the further exploration

of Mars. Within a few months, major boosters of the

MRSR

mission were disassociating themselves from the study and the politically unpalatable price tag ration of

it

had generated

for the mission.

Was

surface explo-

Mars dead?

Another

NASA

facility

—Ames

—stepped into

Research Center

the

new series of missions, dubbed the MESUR. MESUR would be a set of

breach with a proposal for an entirely

"Mars Environmental SURvey" or sixteen to twenty landers that

would form

blanketing the entire planetary surface.

and weather information, enabling

a

network of science

stations

The landers would collect seismic

scientists to construct a global

model

of Mars. The cornerstone of the mission set would be a small lightweight landed station which would be replicated

be so small sent to cle.

in

Mars

Given

volume and mass

at

this

once on

approach,

many times. The landers would

that, at least as

proposed, four could be

a single, relatively low-cost Delta

Ames

complished for less than $1

estimated that

billion.

tag after the sticker shock of

all

of

II

launch vehi-

MESUR could be ac-

This was an impressively modest price

MRSR.

SOJOURNER

66

The proposed

MESUR might not be

a JPL mission, but

it

could yet prove to

be an opportunity for JPL technologies. JPL would show the Mars science

community and NASA headquarters most

deliver the best,

Space Science and Instruments at

JPL

for

weekly lunch

technical guru,

that the

who was

at JPL

One of them was

discussions.

new

using

leading planetary geologists.

The

The head of

a micro-devices

fabrication techniques to create in-

And Dave

small lightweight

that could

the Office of

brought together three key people

credibly tiny instrument packages. Matt

microrovers.

Lab was the place

exciting surface science.

Golombek was one of

Miller

the Lab's

was carrying the banner of

MESUR lander would have room for

only an equally diminutive payload. To capture the backing of the scientists,

would have

that payload

ence. These conversations,

to be capable of performing exciting

which soon included Brian Wilcox, led

sci-

to a

concept for constructing and operating such a payload, not just promising it

on paper. They would build

instrumentation per, soil scoop,

rover,

it

and

would a

a rover massing only a

few pounds, and the

carry: a camera, point spectrometer, rock chip-

micro-seismometer only a few inches on a

side.

operated from a simulated Earth station, would perform a com-

plete science mission.

It

would emplace the micro-seismometer on the

The chipper would wear away

surface, then traverse to a rock.

the rock's

outside surface, revealing the pristine material underneath (which

much greater interest to geologists than the ter taking

rover the

The

outer "weathering rind"). Af-

spectrometer readings of the rock to assess

would scoop up

a small soil

was of

its

composition, the

sample and deliver the sample back to

mock lander from which it had originally descended. Funds from a va-

riety

of sources would be pooled, including JPL's scarce discretionary

moneys, existing rover research

dollars,

and an additional infusion from

Headquarters, together eventually comprising the approximately $1 million necessary to

the

demonstrate

a

new

rover capability.

It

would be

called

Mars Science Microrover. Initially,

Golombek was given an

the effort started. At

first,

allocation of about $300,000 to get

the technical approach

was

and Miller were both given access to the funding, but

unclear: Wilcox little

direction.

The

Golombek was

Little

Rover That Could

manager. With no one supplying the

a scientist, not a task

from the

technical leadership

top,

6?

Wilcox

technical approach this time,

and proceeded to implement

from below He chose it

in a quick, cheap,

a

and

not so dirty way. Perhaps others would follow his lead, perhaps

MSM had no chance of having anything to show bv the next sum-

not.

But

mer

unless they got

moving soon. Wilcox's plan

giving his Robotics group a microrover of Miller's adaptation of

To keep

Rocky

its

had the benefit of

also

own

to

down. Wilcox figured you could

off the rover entirely. "It didn't really matter

the computer, so

you might

as well

put

on the

it

demonstration. The lander was going to have

anyway, so the rover could take advantage of

mostly

a radio-controlled car

tuators

thrown

compete with

3.

the weight of the rover

computer

the

had taken on

into a role he

fell

several times over his career: providing leadership

its it.

leave

where you put

lander.'' at least for this

own computer onboard

The

rover

would then be

operated by the lander, with a few more

in to operate a

sample scoop, and stereo

TV

ac-

cameras to

send back pictures.

One

of the bugaboos of ofF-roading robots was the danger of flipping

While

over.

impending

a rover could tip-over,

and end up on

gates to the public for an

The its

would

would be

wheels to get out of

its

back.

a

'self-righting,"

somehow

jam. Wilcox had

might be able to handle terrain series.

The

ing together with

Timothy Ohm.

it

full

which had

its

hicle

was

a gifted

own motor

360 degrees. The rear

"fork" with wheels

of an

vehicle wouldn't slide its

stilts

putting

concept

in

itself

mind

back on

for a four-

than Bickler's six-wheeled

rolling

around on

stilts.

mechanical engineer

"stilts"

drive

"

that thing rolls over r

ended up upside down, and

Wilcox designed and

wheeled mobility system. The front axle,

a it

as well or better

rover looked like

also an expert machinist.

same

itself

Whenever JPL opened

"What happens when

wheeled rover that wouldn't care whether

Rocky

warn

open house, someone looking over the rovers on

inevitably ask,

ideal rover

sensors to

tilt

you could never guarantee the

off a particular rock

display

be outfitted with

Work-

who was

built "Go-For." a fork-

were both mounted

to the

and could be rotated through

a

were similarly locked together, creating a

on the ends of the prongs. The

could actually do slow somersaults, turning

result

itself

was

over

that the ve-

when

neces-

SOJOURNER

68

sary.

And by

shifting the forks

could be transferred from

its

through smaller angles, Go-For's weight

front to

its

rear wheels.

To go over

a rock,

Go-For could lean back, taking the weight off the front wheels, making easier to

lift

the front onto the rock, while simultaneously putting

it

more

weight on the rear wheels, giving them more traction. Once the front wheels had a good purchase again, Go-For could lean forward, ing

it

easier for the rear

Miller

wheels to climb the obstacle.

was extremely unhappy with the mere

While Wilcox was marketing Go-For by carrying it

now mak-

around, Miller lobbied against

its

existence of Go-For.

it

and driving

to offices

use on Mars Science Microrover.

One day Golombek instructed Wilcox to stop charging to the MSM account. Golombek told Wilcox that a new task manager, Arthur "Lonne"

MSM; Lane would

Lane, was about to take the reins of

decide what the

proper next steps would be.

Lonne Lane and

his skunkworks-style

team had

just

come

off of a

task called Delta Star,

which had produced

finish in only fourteen

weeks. Lane seemed to be a good choice to pull

flight

hardware from

start to to-

gether the disparate elements of a complex rover system on a tight schedule. After

all,

hardware;

it

dience,

the

MSM rover did not need to meet the standards of flight

only had to appear plausibly flightlike to the appropriate au-

and function properly on Earth. As

mass of eighteen pounds trary:

The

mass.

If

for the

a goal,

MSM rover.

The

Lane selected

decision

a total

was not

arbi-

MESUR lander could probably afford to carry a payload of this

Lane succeeded in shrinking the rover as planned, Rocky 3 (which

had grown to forty-four pounds) would look

like a

clunky dinosaur in

comparison. In

September 1991, when Lonne Lane began leading the

he had no experience whatsoever with any of the rover engineers tion system. entist

who would be

Lane did bring with him

who had

rovers.

knew how

a rarity of experience:

He

Lane nixed Wilcox's Go-For concept. at least

He was

a sci-

understood the relevant

to run a flight project development

team, neither of which the rover researchers

where the computer was,

not worked with

key to building the demonstra-

delivered flight hardware.

science community, and

He had

MSM effort,

It

knew much

about.

turned out that

when you had

it

did matter

sophisticated science in-

The

Little

Rover That Could

69

struments on the rover. You needed the computer to operate the instru-

ments and process

most out of

its

their data.

Go-For had the disadvantage that to get the

mobility capability, you needed

some kind of

active con-

monitored where the weight of the vehicle was

trol that constantly

shift-

ing and modified the fork angles accordingly. Rocker-bogie vehicles were generally

more

Lane got

stable,

with no added onboard "intelligence" required.

a crash course in rover

mechanics from

Don Bickler. A group

of mechanical engineers was tasked with creating the next in the series of rocker-bogie vehicles.

The Mars Science Microrover,

have about the same wheelbase as Rocky

3,

Rocky

alias

would

but would weigh a third

as

A common method to get weight down in structural members is to

much.

— a hollow rectangular beam can be almost

make them hollow in

4,

most

cases as

its

as strong

solid counterpart, while containing only a fraction of

the mass. But manufacturing hollow structures of unusual shape can be costly

and

difficult,

and

MSM money was tight.

Faced with the challenge

of keeping both weight and cost low, Bickler would end up inventing a

new machining technique. In the competition for Miller's

station

group got the gold and the simulated

division of labor,

who would

build the rover "smarts," Dave

ring. Wilcox's

group would create the control

lander.

decision

was presented

making the best use of the cadre of rover

reaction, however,

as a simple

engineers.

My

was one of dismay. Our group had been building and

controlling rovers for years, and mately, but

The

knew

now we had been relegated

the relevant technical issues to a support role.

Rocky

4

inti-

would

be directed by Miller's 'behavior control.' Would we be shut out from

now

me not to worry: We had gotten the better, more

chal-

on? Wilcox advised

lenging assignment. ble for

The Robotic

commanding

would build

Once

Vehicles group

the rover, even

if

would

still

be responsi-

the Robotic Intelligence group

the onboard control system.

again, Brian

Cooper would be responsible

cle control station. Like the

Army-funded Robotic Technology Test

cle,

and the Blue Rover before

this

time there would be a

it,

new

rover.

But the

Vehi-

MSM called for a version of CARD.

But

challenge. In each of the previous imple-

mentations, the stereo cameras had been

moved with the

for designing a vehi-

mounted on

the vehicle, and

MSM rover would be only about a foot

tall.

SOJOURNER

70

The

single

camera onboard would be so low to the ground

would hardly be a

few inches

might

tall

totally

block the view The

almost certainly have a camera anyway, and

of a

tall

that

able to see any distance ahead to plan a path.

mast, with a

it

Cooper

A rock just

MESUR lander would

would be

sitting at the

top

much better vantage point than the rover. When the vehicle, you always commanded relative to the

cameras were fixed to the

"Go

current position, such as to the left." If

rover

you used the

was always moving

would Cooper have tell it

where

was

it

to

tell

to a spot

lander's cameras, relative to the

the

you saw more, but now the

cameras' location. Not only

MSM rover where to go, he would have to

starting from.

And

since the cameras stayed behind

would get harder and harder to accurately designate the

with the lander,

it

rover's position

and target locations

also

one yard forward and three yards

as

it

got farther away.

The microrover

had instruments onboard that would need to be commanded. The

new control station would issue commands to the spectrometer to gather data, to the

pick up

rock chipper to

start

and stop chipping, and to the scoop to

soil.

Others in the Robotic Vehicles group, including myself, designed and built the

MSM simulated lander. We called

actually land;

lander

it

was

would do

a platform that

"simulated" because

it

didn't

performed only the functions

a real

it

to support the microrover operating nearby. Except for

MSM

the lack of wheels, the pair of small stereo

lander was similar in concept to

cameras was mounted on a compact

Robby

pan-tilt

A

mecha-

nism elegantly designed and machined by Timothy Ohm. Electronics

in

the guts of the lander could capture frozen images from those cameras

and transmit them back to Cooper's control

would transmit commands

cated near the cameras ceive data sent a

from the

rover.

way off, and back onto,

a spot for

Rocky 4

seismometer

station.

Ramps

it

modem

lo-

and

re-

to the rover,

the lander, while a trough in the middle provided

to deposit collected soil samples.

arrived,

radio

located front and rear gave the rover

would be tethered

which would communicate the readings goods store

Lane aimed the team

When

the micro-

to the lander's electronics,

to a science display station.

covered the lander's external panels with gold Mylar local sporting

A

foil

We

bought from the

for an appropriately spacecraft-like appearance. at a

mid-summer demonstration. To

achieve

it,

f

The

Little

Rover That Could

each subsystem would have to deliver

part

its

?l

on time. The new micro-

rover chassis, computer, and electronics were due at the end of

Allowing until the end of

1992.

ments together, there would

May for installing and integrating the

still

be two months for

ware and testing with the lander and control looked

.

.

.

possible.

March

Would we be

able to pull

it

ele-

finalizing the soft-

station.

The schedule

off?

# At the

Golombek and Dave

end of October 1991, Matt

Washington, D.C., to pitch the Mars Science Microrover Science

Working Group meeting. This group was

scientists that advised

From

sions.

NASA on

the

Miller flew to

at the Sixth

Mars

team of planetary

science priorities for future

Mars mis-

the science advocacy standpoint, these were the very people

Mars Science Microrover had been designed to impress.

Much science

MESUR mission. The Would MESUR do credible science?

of the two-day meeting focused on the

community was concerned:

For engineering reasons, sampling of subsurface materials was not part of left dust, soil, and surface rock. A potenMESUR was the Alpha Proton X-ray Spec-

the mission designers' plan. That tially

important instrument for

trometer, which

of rocks and

made

of?"

To

was capable of determining the elemental composition

soil,

answering the fundamental question, "What are they

APXS would have to be placed be analyzed. But how do you get the

give meaningful results, the

in direct contact

with the target to

APXS onto rocks? And since you would have brought the instrument all the way from Earth, you really wanted to put it on more than one rock. The chief engineer for the MESUR study at Ames suggested several options for deploying the APXS from the lander, which included dropping the APXS on the surface, catapulting the instrument out from the lander, catapulting

it

and then reeling

it

back

in,

aiming the catapult

first,

nally the possibility of placing the instrument about three feet

lander using a robotic arm. ideas dimly.

They were

The Working Group

scientists

and

fi-

from the

viewed these

so unimpressed with the options that they agreed

to consider replacing the

APXS with

other instruments that did not

re-

quire placement in such close proximity to their targets. Scott

Hubbard from NASA Ames presented

the concept of

SLIM

SOJOURNER

?2

(Surface Lander Investigation of Mars) to the

working group. This was

a

proposal for a "mini-MESUR," a single lander that would be launched in 1996, four or

more

years sooner than the

first

proposed

MESUR launch.

The hockey-puck-shaped lander could land either right side up or upside down: The instruments would be configured to deploy properly either way. The design (and the science) would be simplified to bring the cost within the $150 million limit imposed for NASA's

new

Discovery-class

low-cost missions.

This was the

first

time anyone in the Science Working Group had

heard of SLIM. Certainly nobody ing.

the

Although SLIM would put

at JPL

was aware of it prior to the meet-

a lander

on Mars many years

earlier

than

MESUR mission, the working group wondered, what good was a sin-

gle lander?

The concept was not

received well.

Golombek's microrover presentation was not scheduled until the

sec-

ond day of the meeting. He was nervous, and could not sleep the night before.

This would be one of his

first

presentations in front of a large group

of science heavyweights. "I'm going to get toasted

At

in his head.

3 a.m.,

Golombek found himself

troductory viewgraph for the start of his

talk.

alive,"

was the thought

hand-lettering a

He knew

that he

new

in-

had

to

MRSR rover study had created MRSR legacy was that rovers needed

break through the perceptions that the within the science community. to be large

and complex to be

The

useful; they

were

difficult to operate, re-

quiring a major infrastructure; and they were expensive.

Morning came. At his presentation, Golombek addressed the negative perceptions head on. His premise was that the belief that surface mobility

required a rover the size and complexity of

MRSR was just plain wrong.

Rovers could be small and simple, easy to operate, and cheap. The just cried out for a microrover to deliver

prove that a small rover could do

it

to rocks.

And JPL was going to

useful science within another nine

months. Mars Science Microrover was already under

Golombek

listed the existing

APXS

way

instrument payload for

MESUR, and

proceeded to describe what a microrover could do for that payload, improving the science you could get back.

A rover could get up close to Mars

rocks, providing the equivalent of a geologist's

many rocks,

hand

lens. It

could inspect

not just one. Rockets firing during landing tended to create a

The

Rover That Could

Little

?3

A microrover could

contamination zone immediately around the lander. traverse outside of this zone, retrieve soil or dust,

the analyzers

on the

MESUR

lander.

and bring

called for the

back to

it

placement of many

seismometers, one per lander. Vibrations from the lander

or the

itself

Martian wind could corrupt the seismic measurements; the rover could

emplace a small seismometer out of the wind.

And

tively explore the

ing, instead

finally,

landing

away from the lander and

safely

a rover

site,

would allow the

science

potentially

team

going out to features that looked

to ac-

interest-

of leaving them tantalizingly out of reach.

Dave Miller followed Golombek with small rovers.

He showed

a talk

about the

feasibility

a videotape that demonstrated Tooth, the four-

wheeled tabletop robot about twelve inches long. Tooth looked like but

perform

actually could

it

a

away from such

front,

a light.

one that would close

Tooth

if

a toy,

crude mini-mission. Photodetectors on-

board could determine the direction of a bright drive

of

also

light,

had a very simple gripper on the

any object blocked the

on the

gripper jaw from the light detector

would

so Tooth

other.

light source

You could

stick

on one

your

fin-

ger between the gripper jaws and they would close on the finger. But once the gripper fully closed,

Tooth was

it

would

after bigger prey: the plastic

cap to a spray-paint can.

gripper could only close partway, Tooth grasp.

knew it had what

So Tooth would wander away from a

propriately sized plastic cap.

and

reject the finger as too small

Once

it

light until

it

it

let go.

When the

wanted

in

its

ran into an ap-

had grasped its "sample,"

its

behavior

Now it was enamored of light, and headed back home toward the light source. When the light source got bright enough, Tooth changed changed:

its

set

mind

it

again, deciding

down

it

didn't

want the cap anymore. The

the cap by the light, then

Given enough time and caps in the ber of caps and deposit them

all

MESUR

seemed

clear. Miller

near the

Tooth could

light. If

collect

caps.

any num-

you imagined the

light

the caps as small rocks, the relevance

had brought Tooth with him, and he

demonstrated some of Tooth's simple behaviors, including to grab hold of his finger

robot

wandered away, seeking more

vicinity,

source as a point on the lander, and to

little

and then

its

proclivity

politely let go.

By the end of the meeting, the Science Working Group's recommendation to

NASA was

to

add a microrover to SLIM, and go

for a launch in

SOJOURNER

74

1996

The rover way to deliver

the funds could be found for a "quick Mars mission."

if

would make

a single lander like

SLIM

useful,

and provide

a

APXS and potentially other instruments to their science targets. On the way home, Golombek was almost literally jumping up and down. Before the Working Group meeting, JPL had been pretty much cut out of future Mars exploration. Now it had the hot technology that made the

Mars mission worth doing.

a '96

Within a year of Ames's original

months

after the

MESUR

proposal, and only

two

Mars Science Working Group meeting, JPL became the

NASA center for the new Mars Exploration Program, and the place where

MESUR would be implemented. proved,

NASA

became

real, to JPL.

While the mission had

had handed the charter

to

do

it,

if

still

not been ap-

and when the mission

# The date of the Mars Science Microrover demonstration had only been

vaguely defined as middle to

summer. The

late

objective

had been to

de-

velop a compelling microrover capability in time to influence the funding cycle for the next fiscal year.

Then, near the end of

a

new

directive: the

midst of

Rocky 4 mobility

dealing with delays in the completion of the

Lane received

April, in the

chassis,

MSM rover was to be demonstrated as

the centerpiece of the celebration of the twenty-fifth anniversary of the first

day,

Surveyor soft-landing on the surface of the Moon, to be held on

June

26, 1992.

There was onto.

The

still

Fri-

That was barely two months away!

no rover body

MSM group,

to install the

computer and instruments

already working hard,

was pushed

into a frantic

mode. Lane began holding meetings every morning. The group would semble

in Building 107, standing

as-

around while Lane or the system engi-

neer outlined the schedule for that day. The mechanical team

finally

delivered the rover chassis with flightlike stainless steel wheels in May,

over a

month

group get

its

after originally

promised. Could the Robotic Intelligence

computer and software

installed

and working

in the six

weeks remaining?

The contracted-out computer job any

easier.

To

fit

electronics boards did not

into the limited

volume

available

make

their

on the Rocky 4

The

chassis, the rover brain side.

The

?5

cube a few inches on

as a

a

were prototype boards with the necessary

components mounted

manufacture proper

Rover That Could

had been envisioned

walls of the cube

electronic

Little

to them.

There wasn't enough time to

circuit boards, so the circuits

would have

to be "wire-

wrapped" with individual wires leading from each component to the next.

With hundreds of wires

would look

computer cube

crisscrossing over the boards, the

Lane

like a bird's nest.

didn't feel that Miller's

group had the

time to build the computer themselves, so to do the job he had selected a local

when

company

often used by JPL.

the cube arrived,

tightly

it

was

They

The wires had been wrapped too component pins at the point where

a mess.

and tended to break off of the

the exposed wire

many

available in

met

its

insulation.

Worse, although wire-wrap wire was

you could

colors so

usually did excellent work, but

tell

one from the other

in just this

type of situation, the contractor had used only one color. Debugging the

boards would be nearly impossible! But they did

it

anyway.

As the hardware and software integration raced forward, the of the demonstration began to core of the

festivities,

Lane's

come

together. Since the rover

site.

went up around

(Although both rovers and rockets had been tested in the

the Arroyo Seco itself was not city

at the

MSM team had been handed control over the

orchestration of the event. Permits were sought. Fences

the

logistics

was

vicinity,

on JPL property, and in fact belonged to

the

of Pasadena.) To acquire "descent imagery," showing the equivalent

of what a Mars lander might see on

its

way down, Ken Manatt, one of the

engineers on the team, flew his hang glider from the top of neighboring

Mount Wilson to JPL, snapping pictures

as

he maneuvered over the rover

operations area, and finally landing in the debris basin just south of the

Laboratory Before the team was ready, Lane forced the group to tions to the

outdoor demonstration

site.

move

He knew that the

its

opera-

system had no

chance of working unless the team was familiar and comfortable operating outside well before the big day.

The Robotic Vehicles group and the Robotic Intelligence group had culty collaborating.

From the

diffi-

days of Robby, the AI team viewed the rover

SOJOURNER

?6

navigation testbed as a behemoth, and the approach of modeling the rain

around the vehicle

in detail as misguided.

Robotic Vehicles group, the

common

exasperated "They don

it!"

ated the other

t

get

members of

When

talking about the

refrain within the

One by one,

ter-

AI team was an

their arrogant attitude alien-

MSM team. Now that the AI team was in

the

the crunch to deliver a working rover, they were not so confident. This

was

their opportunity to

ior control," ideas, for

some

it

was

also a chance,

to

fail.

Some

but

them

ff

demonstrate the

reality

of their vaunted behav-

independent of the veracity of their

worked didn t, and

things that should have

things they just didn't have time to

do

in the first place. Readings

from simple sensors were supposed

to trigger behaviors that led to a safe

traverse that avoided rock hazards.

They had mounted small sensors on

the corners of the rover that radiated infrared light and looked for

of that light to be bounced back.

would

frared, this

and needed to veer to

many

indicate that off.

If a

sensor detected

Rocky 4 was approaching

But the infrared

light detectors

There was just not enough

rocks.

some

the rocks were effectively invisible, there

light

a

some

reflected in-

rock hazard

seemed

to

be blind

being reflected back.

would be no sensor readings

If

to

new behavior, and Rocky 4 could bump into things. There was no time to come up with a new sensor type, so the only strategy left was to move enough rocks out of the way to leave a clear path for the rover. trigger a

The Robotic

Intelligence

group was under extreme pressure

and would sometimes respond to that pressure by blaming

to perform,

their

woes on

others, legitimately or otherwise.

The demonstration site was located in the Arroyo Seco just>east of the Laboratory.

The

MSM team created a miniature rock field by carefully ar-

ranging rocks of several sizes brought in for the purpose. Large opensided tents were erected nearby so that the

would be

able to

trol center

sit

comfortably out of the

was located within

VIP audience and media

summer

a large rental truck,

sun.

During

we would often

rover con-

with shock-mounted

computer workstations and video monitors strapped testing

The

leave the rear roll-up

to the inside walls.

door of the truck

open, but during the actual demonstration the door would be shut, forcing the operators to rely only

on lander camera images and video from

The

Little

Rover That Could

??

the rover-mounted camera. In addition to the rover control computer, an-

other workstation was dedicated to processing and displaying the science data from the micro-seismometer and the rover's onboard spectrometer.

Cables ran from the computers inside the truck to video monitors in the

audience viewing area. Outside, the simulated lander had been precisely

one end of the rock field,

situated at

its

ramps leading down

into the con-

structed terrain.

Soon, another lander stood nearby: JPL's full-sized model of Surveyor

had been

carefully trucked out to the

Arroyo

site

and situated just outside

the rover's test course.

On

the day of the demonstration,

the rover control station.

Months

I

found myself

before, Brian

sitting in front

of

Cooper had planned an

expensive vacation with his wife, before the demonstration date solidified at just the

cation

wrong time. Cooper had nonrefundable

won

.

.

.

and now I was

backseat driver

who would

in the

hot

airline tickets.

The

va-

Brian Wilcox sat behind, the

seat.

have time to think while

I

was busy typing

commands. In the

week

or so before, Rocky 4 had successfully performed

the steps of the demonstration.

To get more

floodlights to illuminate the test course night.

The

trickiest part

test time,

and often operated

you directed the rover to

tance to a desired spot, slippage of the wheels in the it

to

the rover's

end up

set

in a slightly different place.

Looking

drive

soil

its

target,

and then move

it

up

so in

some

dis-

or on rocks

at the

video from

own camera, you could estimate how much you'd have to

the rover to face

of

late into the

was getting the microrover positioned just

front of a target rock. Every time

caused

we had

all

turn

forward to put the chipper

against the rock surface.

Our cials

usually deserted test area

had been bussed out to the

was now

site.

bustling.

JPL and

NASA

offi-

Television crews and press reporters

had come with them.

The crowd followed Lonne Lane over toward Rocky 4 had been placed on the top of the

the simulated lander.

lander, facing backwards.

rock chipper was a robotically operated and aimed the front of the vehicle.

The chipper was now

tilted

ice

Its

pick sticking out

upward,

as if to

warn

SOJOURNER

?8

away the from

a

might threaten the

inquisitive giants that

hook slipped over

the chipper

was the

little

rover.

Dangling

tiny but functioning micro-

seismometer.

By command from

inside the truck,

By glancing over

the rover's camera.

I

turned on the seismometer and

at the

video monitors, the audience

could get a rover's-eye view of the events to come. Certainly Lane was explaining this and

much more

to the assemblage, but

I

was oblivious

to

The signal came and off we command, which was relayed from

everything except the signal to start the demo.

went.

I

sent the "Rover Disembark"

the control station to the lander, and thence over the radio link to

And

off the rover went, rolling rapidly

seismometer held high and its

down

ramp onto

the

Rocky 4.

the sand, the

out of the way. The seismometer trailed

safely

data transmission cable behind

it.

Rocky

4's

backwards driving avoided

trampling and potentially entangling the cable. Next Rocky 4 lowered the chipper until the seismometer rested in the

then rolled backwards

dirt,

another eighteen inches, leaving the seismometer deployed where

it

be-

longed.

So

far so

good.

We

points that would get rock, drove a

it

it

then sent the rover on to two designated way-

to the rock of choice.

We got Rocky 4 aimed at the

forward, and tilted the camera-spectrometer platform up for

good close-up of the

Spectrometer data flowed

target.

time to activate the chipper, which would batter its

moving the

tip against

it

was

the rock, re-

surface layer in preparation for getting another spectrum.

The rock chipper kept going.

Now

in.

No

started up.

It

rattled

away

at the rock.

And

it

just

message came back from the rover to confirm the com-

mand. I scrambled to send an "Abort" command.

The

merrily chipping away.

Shutdown" command.

I

only thing

knew

if

I

left

No effect: The rover was

to try

was

the

"Emergency

sent the shutdown, the demonstration

would take too long to re-initialize the rover after that. Wilcox conferred for a few seconds. "Send it." Nothing happened. Rocky 4

was

over:

and

I

It

was no longer responding crashed, and there

to

commands. The

was nothing more we could

rover's

do.

computer had

The chipper ran

on,

oblivious.

The audience began

to realize that the demonstration

ceeding properly. After a moment,

was not pro-

Lane strode over to the rover.

He

The

reached

down and

Rover That Could

Little

was having

ference from

all

The

press

due to radio frequency

inter-

of the news crews' video cameras nearby.

problem during

died.

persis-

Lane calmly explained that the

a problem, quite possibly

Inside the truck, a

The embarrassingly

shut off the rover power.

tent chattering of the rock chipper stopped.

rover

pg

was stunned by the

I

and

testing,

would have

failure.

at just the

We had never had such moment

worst

the rover had

a field day!

Outside, Lane continued to

tell

the story of the rover.

He

asked the

audience to imagine the next steps the rover would follow, extrapolating

from what they had already sion, the

Although temporarily halted

seen.

microrover had already done real science.

in the tent,

you could

see the images the rover

camera. There was the

first

On the

had taken with

The JPL

porters saw the

data set from the mini-spectrometer.

And the

results

even

mission, the rover need only traverse to the

its full

soil,

and return

it

to the lander for

NASA Associate Administrator, as a success. Lane knew how to

Director, the

rover

little

mis-

onboard

nearby sandy area, scoop up some analysis.

its

its

micro-seismometer deployed by the rover was generating now. To accomplish

in

video screens

audience and give them a piece of his vision.

He

later

and the

re-

talk to his

commented,

"It

helped that no one had any expectations of a microrover's capabilities to

compare with." Rajiv Desai,

Dave

perceived fiasco.

Miller,

Once

the

and

their

team were

of their

livid in the face

crowd had moved on, they took Rocky 4 back

to the staging area and attempted, unsuccessfully, to revive the vehicle.

They formulated

a hypothesis as to

what had probably happened: The

vi-

bration of the rock chipper had caused a short in the poorly wired com-

puter board, frying the

CPU chip. The CPU had turned on the chipper just

but there was no brain

fine,

had

effectively

the

CPU

possibly

committed

chip out of

stomped on

its it.

left

when the

suicide. In frustration,

for a test

more thorough

one of

socket on the board, threw

it

it off.

their

Rocky 4

team pulled

to the ground,

and

Desai and Miller blamed the hardware, and Lane's

poor choice of contractor to build

Ken Manatt picked

time came to shut

it.

the chip out of the

evaluation.

environment, the chip

at first

When

dirt.

He took it back to

he powered

would not

operate.

it

up

The

in a

the lab

benchtop

small onboard

SOJOURNER

80

memory, which told the chip how to seemed

when it was first turned on, Once Manatt reset the memory to its factory set-

to be corrupted.

tings, the chip

functioned normally, despite the mistreatment

had not been

it

sufficient to

Lonne Lane's own surmise was ply not been given

enough time

simply crashed. In his view,

available

had

re-

permanently damage the CPU.

that the rover software

to integrate

ware that drove the various onboard anyone to have

it

had been a short in the wiring of Rocky 4's computer

ceived. Perhaps there

boards, but

initialize

fully

between the

and

team had sim-

test the low-level soft-

and the computer had

devices,

might not have been humanly possible

it

for

completed the software job, given the few weeks of the rover chassis and the immutable

late arrival

date of the demonstration.

The

of Rocky

definitive cause

The sweep of

events to

seizure

come would leave

gation, nor a strong need. For

JPL's investment.

4's

With

a

little

would remain

a mystery.

time for a detailed

investi-

Mars Science Microrover had delivered on

margin of seconds, Rocky 4 had survived

long enough. Despite the early termination of the demonstration,

had proven the

viability

just

MSM

of the microrover as a component of Mars sur-

face exploration.

A few days

later,

Donna

Shirley called an all-hands

meeting for the

MSM

team. First she congratulated the team for a job well done. Then she

moved on to

the real purpose of the meeting.

There was going to be for

a

new

mission to Mars.

Mars Environmental SURvey.

It

was

would put many landers on the Martian

It

was

really a series

surface.

The

called

MESUR,

of missions that

first

launch would

demonstrate the technologies that would be necessary to make the entire mission set work;

it

was

sort of a trailblazer for

MESUR,

so

it

had been

named MESUR Pathfinder. (The project had no connection with the planetary rover research

seemed

to

program

be a popular name

that in

had produced Robby. "Pathfinder" just

NASA

circles.)

The mission had not

yet

been approved, but that would come soon enough. Meanwhile Code R, the

NASA organization that among other things controlled NASAs fund-

ing of automation and robotics research, was going to pay for the devel-

The

opment of tended to not fund

Of

8

Mars Science Microrover,

MESUR Pathfinder. This was unprecedented:

on

in-

Code R did

flight systems.

course, sending a rover to

Shirley

also unprecedented.

to be the rover project manager.

months she would be assembling the

huge amount

"pony up" $25 million

to

for a research task, but

new

flight

team, and

Over

many

room today would participate.

Code R was going tem. The

Mars was

announced that she was

of those in the

flight

rover

and was

officially

it

That was

for the rover.

was not very much for a

team would be

lean.

It

a

flight sys-

needed to get started

MESUR team. When MESUR did get its act to-

now, to get a jump on the

ity,

Rover That Could

a flight microrover similar to the

fly

the next few

gether,

Little

approved,

it

would be

a significantly larger activ-

moving much faster than the smaller microrover team could ever hope

The

to do.

flight rover

would have

would fact,

to build

up

inevitably catch

team would need

up with

it,

the rover

MESUR

the people planning

a

own momentum

its

head

start.

would not be

I

and the rover would be

had held that

secret

I

some kind of pay-

hope even I

as

I

first

came

to

how

to

make

rovers work, but

seemed

work

had come to accept

a real mission would never happen. Each research task taught us

about

the

fly

had watched and read about

But over the years of working on rovers,

at JPL.

behind. In

a mission that explored space, like

the Mariner, Viking, and Voyager missions I

left

MESUR

ready.

had always wanted to be part of

while growing up.

rover effort

had not yet even promised to

rover onboard their lander. But the lander had to carry load,

The

soon, so that while

to lead only to

that

more

more

re-

search.

Only

flight rover

canceled

moment,

at this

become

real to

as Shirley described the plan, did the idea

me. There were

—or never approved

personally that deadlines,

I

would be

and money,

all

of a

still

doubts: Missions could get

in the first place

—and no one had told me

part of the flight team. But there were plans,

aimed

word "FLIGHT" was stamped

at building a

all

over them.

new

microrover, and the big

®L

-0-

:••-•

H

PART

2

m PATHFINDER

SEVEN

SMALL ENOUGH TEAM TO DO THE JOB

A

To

begin work

on

a flight project

is

to enter a

new world where

mass,

power, and volume are precious commodities. Consuming too

much

of any of these

aerospace they

call

is

—

not an option. Each available rocket

them "launch

vehicles"

in

—whether a Delta, Titan, or

Ariane, has only so

much weight of payload it can put into a particular tra-

jectory in space.

you

If

are launching a spacecraft to Mars,

too high, the laws of physics ensure that spacecraft

must carry with

it its

it

own power

and

its

mass

is

target.

Each

source, whether in the

form

will never reach

its

of solar arrays, radioisotope thermoelectric generators, or batteries.

These power sources are limited spacecraft that

depend on

this

in their capacity.

power must use

it

The components of the efficiently for

needs of the system exceed the available power, the spacecraft each spacecraft has to be small enough to fairing, the

nose section that protects the

namic drag during the

fit

when the dies. And

within the launch vehicle's

spacecraft:

rocket's supersonic flight

and reduces aerody-

up through the

Earth's

at-

mosphere.

None of tasks,

these commodities had been critical in the rover research

where the usual

final

product was a videotaped demonstration. And

while the Mars Science Microrover had been designed to appear

flightlike,

— SOJOURNER

86

that vehicle only

had to operate

for

about an hour under

own power

its

while surviving the very Earthly environment of the Arroyo Seco.

The

microrover would be

flight

the Pathfinder spacecraft were trickling

The

flight rover

would have

to

The

different.

fit

down

imposed on

constraints

to each payload element.

within the tight confines of the

Pathfinder lander, survive the rough trip to Mars, and operate

where the surface temperatures ranged from

a high of

low of — 130°F. Somehow the rover would have

power

to run

hours.

The

its

rover

computer and motors

would have

to

on

a planet

about 60 °F to a

to supply itself with

for days or weeks, not

do more than merely

survive:

It

merely

would

unknown set of science instruments, and perform an as yet unknown mission. The whole package must fit into whatever mass allocarry an as yet

cation Pathfinder could spare. This

would probably amount

to

some ten

or so pounds. It

would be

Donna

Shirley

a lot to stuff into a small box.

was

a

manager with no one

to

manage. Only

a

few of the

engineers on the former Mars Science Microrover team had the proper skills for

the flight effort.

the position of Rover

be

First

efforts Shirley

had achieved

Team Leader, but building the rover team would not

as simple as calling

work.

Through her own

up the people she had

she needed to identify

who

gotiate their availability, convincing

in

mind and putting them

to

the right engineers were and ne-

them to leave their current JPL jobs to

join the rover team. Little

Shirley's

make

money would be

available in the first year.

But

team would plan the schedules, determine the budgets, and

the key design choices to prove the feasibility of delivering a flight

rover on-time and within the $25 million cost cap.

the "burn rate"

—would jump

The

rate of spending

in future years as the detailed design

implementation phases began; but there would be no more money the $25 million tion

time

in that

on

was gone. This

the rover: If Shirley's

total cost

if

necessary, until they

after

was NASA's number one condi-

team came up with

a design that

then they would go back and simplify the rover, making slower

and

had an acceptable

it

exceeded

it,

dumber and

sticker price.

A Small Enough Team

Shirley

wanted lead engineers

to

Do the Job

8?

for each of the rover subsystems:

Power, Telecommunications, Mobility-Thermal-Mechanical, Control and

good people and she needed them

Navigation. She needed

right away.

She went to Charles Weisbin, manager of the Automation and Control section. Shirley

for

all

wanted one "Cognizant Engineer"

what

be responsible

of the control and navigation area. This subsystem would provide

the rover brain, the software that

would

to

rely on,

to do.

would run on

and the ground control

Did Weisbin have anybody

it,

any sensors the rover

station that

in

would

—there had been no prior robotic

came which Henry

flight rovers

flight experi-

—so the question be-

of the robotics research engineers in the section

best suited to

move

the rover

mind?

There were no rover control engineers with previous ence

tell

would be

over to the flight side of the house. Weisbin selected

Stone, another

member of Wilcox's

was then the manager

Hazbot

for the

Robotic Vehicles group. Stone

task, in

which he had proven

his

mettle by adapting a commercial robot platform to operate in hazardous

He had

environments on Earth.

also

shown

a

knack

for digging deeply

into the details of whatever project he focused on. Weisbin thought Stone

could well apply those

Stone jumped

traits to

the flight rover.

When

Weisbin asked,

at the opportunity.

Stone was impressed with

how

both Donna Shirley and Pathfinder

"were putting together the project in a completely different way. There

were no

rules, or if there

ho bunch of

were, they were to be broken.

a part of

and Navigation subsystem.

had

for the

He

sultation with Brian Wilcox, I

a

gung

it.

Soon Stone was hunting

me twice.

was just

who were going to make that thing happen."

rebels out there

He wanted to be

It

system engineer for

did not

his

new

Control

know me well, but after some

he came to

con-

my door. Henry didn't have to ask

my place on a Mars rover flight project.

# Howard Eisen joined the team side.

as Stone's

counterpart on the mechanical

Since his days as a co-op student experimenting with

Robby Jr., and working with had completed

his

degree

at

Bickler

Robby and

on the development of Rocky, Eisen

M.I.T and returned to the JPL Mechanical

SOJOURNER

88

Systems division

two JPL

as a

permanent employee. He had already participated in

flight projects for Earth-orbiting satellites.

only in his mid-twenties, but he had already

He was

shown himself

mechanical engineer with a knack for getting things done.

made up

brash and abrasive, but he

and an

intuitive grasp

for

it

young,

to be a gifted

He

could be

with creative engineering

of the relevant technical

still

skills

He and Shirley got

details.

along famously.

Shirley continued building

her team. Upper management told her the

name

more person who must be on it: Bill Layman. Layman was a veteran engineer, one of the most senior and respected

of one

at the

Laboratory.

He had

craft that, except for Pluto,

Layman

personally designed hardware for

had explored every planet

JPL

space-

in the solar system.

had, over the years, gained a wide reputation for being an un-

He

prejudiced mediator.

had been going nowhere

were

turf wars

once recalled being asked to manage a task that

due to inter-organizational squabbling. "The

fast

the high

all at

mucky-muck

level,

so

I

just gathered the

troops around and basically said, 'Here's this really hard problem. Let's figure out

division

got the

how we

can solve

And

people pretty quickly forgot which

numbers they had stamped on their foreheads

name in some

ing a neutral party. as I've got

division managers' eyes,

I

... In the process

I

think on both sides, as be-

don't really care a lot about division politics so long

I

an interesting job. So

visions decided that they

ing

it/

managed by Donna

I

think,

based on that experience, the

needed someone

to

di-

add to the brew that was be-

Shirley."

Layman and Shirley talked about how they should split up the job between them. "We both tried to outline where we thought we were strong, what we might be able to bring to the party," Layman recalled. "We agreed that there was plenty of work for the two of not write

work as step in,

down

it

us,

why didn't we just

a role statement for the next few months, and just take the

came.

and vice

When one person was too busy, versa.

ing into a nightmare

You always worry about

—but

it

actually

the other person

a contract like that turn-

worked out just the

The other engineers working on

would

the rover effort

opposite."

knew nothing of

A Small Enough Team to Do the Job

89

management's imposition of Layman onto the team.

Layman to

the core

team

as a great

Shirley introduced

mechanical engineer, perhaps the best

made it clear that the team was lucky to have him as the microrover Chief Engineer. The team members who had worked with Layman in the past knew that was true; the rest of the engineer of any type at JPL. Shirley

team soon learned Over time the Shirley

had

supreme

a

themselves.

it

roles of Bill

hand

Layman and Donna

in fighting the managerial

every project.

Bill

Shirley

became

clear.

in the technical design of the rover, but she reigned

Layman was

and

political battles that swirl

around

the engineering problem-solver and the

Someone on Layman with the bad news,

source of technical leadership and vision for the rover team. the rover

team put it this way: "Well, we go

to

and Donna with the good news."

who would build the

As the team of engineers

flight

microrover began to

assemble, Dave Miller and his Robotic Intelligence group were nowhere to be found. In the fight over

section

had

Intelligence

lost.

who would

control rovers at JPL, Miller's

During the Mars Science Microrover

effort,

the Robotic

group had been transplanted into the same section

botic Vehicles group. Miller and Desai

ment, and by the end of

When the

pointing out that the

hostile to their

as the

Ro-

new manage-

MSM were looking for a way out.

charter battle

been directed to move

were

its

was

lost,

the Robotic Intelligence group had

offices to Building 107. Miller

facilities in

and Desai balked,

107 were not nearly as plush as their cur-

rent offices. In response to Miller's complaints over the unsuitability of

the building, JPL

management broke

the existing cubicles

But Miller and

return. Rajiv Desai

crew would never occupy those new

dry, Miller

had gone on

sabbatical

offices.

Many

their old section,

offices.

By the

from JPL, never to

and the other members of the Robotic

group migrated back to rover business.

funding to replace

on the second-floor "penthouse" with walled

his

time the paint was

free construction

Intelligence

promising to stay out of the

of them would have been accepted into the

they had wanted to be part of

rover team,

if

Intelligence

team consisted of researchers

it.

flight

But most of the Robotic

intent

on proving out new

al-

SOJOURNER

90

gorithms.

They

lost interest after the basic principles

had been demon-

strated.

Yet Miller's group had mission. for the

sonally

The team had developed

real contributions to the

Mars rover

the software and control electronics

Mars Science Microrover. Perhaps most important, Miller had per-

promoted the idea of small

NASA, and

JPL,

made very

the science

robots,

and redirected the focus of

community away from approaches

that

were

too grandiose to be realizable implementations for Mars. Without the

new

microrover mind-set, there would certainly have been no rover on

MESUR Pathfinder,

and perhaps no Pathfinder mission

And several engineers in the

July of 1992, Shirley

Monday morning. JPL rover

on the second

floor of 107.

began holding weekly "core team" meetings, every

Early on, the attendees included the usual suspects of

research:

Don

Bickler, Brian Wilcox,

and other group supervi-

sors with experience in the relevant technology areas.

nizant Engineers for each of the subsystems

supervisors faded into the

mate

all.

Robotic Vehicles group were very happy

to accept Miller's legacy of walled offices

In

at

Then

the Cog-

came on board, and

the

woodwork. The Cognizant Engineers had ulti-

responsibility for the success of their subsystems. In the

ronment, where most employees worked multiple tasks in

JPL

envi-

parallel, often

charging to four different accounts, the Cognizant Engineers would devote their

full

energies to only one job.

The Cognizant Engineers would

much as possible,

staff their

own subsystem teams. As

they would try to bring engineers onto their teams

time. But their budgets to be done. Nearly

were limited and there were many specialized jobs

two hundred people would eventually

the design and development of the flight rover. Yet even effort

was

in full swing, there

thirty full-time

employees on the

all

sorts of

when

the rover

payroll.

was up

for grabs. Shirley's fledgling

new rover, just as her But we couldn't tarry,

technology options for the

old Mars Rover Sample Return team had done. since

participate in

would never be more than the equivalent of

In the beginning, the rover design

team studied

full-

we were on the hook to

deliver a

working Mars rover

in three

and a

A Small Enough Team to Do the Job

half years. "First

your

was

it

like

going through a catalogue and picking out

Layman. "There were

favorite goodies," said Bill

others

you couldn't hope

to

all

kinds of things

way from vision and mapping, Some of them you could adopt completely; adopt given the constraints we had for size,

the technology tasks had developed, robotic control, to mobility.

91

the

all

power, and the schedule."

One

of the big questions was

"What

brain for Mars Science Microrover

would not

Rocky

4's

rover's

CPU must be

people

who wanted the rover's computer to be really smart,

ful.

The

look

will the rover's brain

A

do.

group supervisor named Leon Alkalaj. Alkalaj looked

Mars

real

There were

"flight-qualified" for the rigors of space.

lead proponent of the "egghead" rover brain

like?"

really

was an

power-

electronics

at the flight

rover as

new computing technologies: advanced proces-

an opportunity to develop

multitasking operating systems. Flight projects were where the big

sors,

money

was. Pathfinder's rover should have the resources to implement

the high-performance computing research Alkalaj had been pursuing.

Layman the As Layman recalled,

Soon he learned from cost flight project.

Bill

"I

Leon was the key the thing should care

I

low-

had by that time become well

enough acquainted with how much power was spacecraft designs

new

real constraints of this

available,

and

in previous

had run into the problems associated with speed

player at that point.

I

basically put in front of

draw no more than three-quarters of

a watt,

him

and

I

.

.

.

that

didn't

how dumb it was, that we would just slow down the rover as much as

was necessary

would be telling

to

make

it

think slowly. As long as

okay." Three-quarters of a watt?

him, "Either burn

well. Alkalaj

had no

my

strawman or adopt

little

would be no bigger than Rocky fit

4,

size

To

find a

into the past.

it

it." It

didn't

burn very

both survive the space

of the rover. The

flight

and would be solar-powered. The

on top of the rover would generate only

few watts of power. Those watts would have to the rover's brain, but also for

clearly,

power.

Layman's constraint derived from the

small solar array that could

would think

That was nothing! Layman was

alternatives to offer that could

environment and draw so

rover

it

its

be enough not only

radio, sensors, instruments,

CPU to meet this severe power constraint,

the

a

for

and motors.

team reached

The 80C85 microprocessor was low-power and

rock-solid in

SOJOURNER

92

the face of the radiation environment of space and Mars.

was already about twenty years

sign

old.

It

The 80C85

was slow, with perhaps

the speed of a typical desktop computer then available. But

what the rover needed

it

de-

a tenth

would do

to do.

What if a better CPU showed up sometime later? To keep the

options

open, the electronics that interfaced to the onboard sensors and motors

would be

built

swapped out

on

if a

CPU board could then be

a separate circuit board; the

new

alternative

became

available.

Unlike most JPL spacecraft, built with duplicate components for ability,

one

radio,

would be

the rover set

"single string,"

with only one rover brain, one

of sensors. The usual deep space project would employ en-

gineers to analyze every possible

quence of the

failure

way

the mission could

fail

two

suring that only the simultaneous failure of at least kill

as a conse-

of a single spacecraft component. The project

engineering team would then design out these "single-point

could

reli-

the spacecraft. This approach

failures," en-

related parts

made JPL spacecraft very reliable.

But it also made them bigger, heavier, and more expensive. Pathfinder and its

rover were supposed to be "faster, better, cheaper."

a fully

new

redundant rover:

single-string rover

It

wouldn't

would

fit

and

it

would

We couldn't afford

cost too

survive only as long as

its

much. So the

weakest com-

ponent.

During the summer of

'92 the rover

team struggled with

major design questions. Should the rover be

six-

all

of the

or eight-wheeled? Should

the onboard batteries be rechargeable by the solar array or not?

would we keep the Martian night?

electronics

What

and

sort of radios

batteries

from

would be

How

freezing in the cold

best to

communicate be-

tween the lander and the rover? What kind of cameras should be

mounted onboard? Would

there be cameras at

all?

Should the rover use

CARD or behavior control? How should the vehicle detect and avoid hazards? How much improvement in navigation performance over the research rovers

would we need

What was the

to satisfy our mission objectives?

microrover's mission to be? Just as

purpose was to prove that the landers

was

feasible, the rover

would show the

utility

MESUR Pathfinder's

MESUR

mission concept of small cheap

would be

a "technology experiment" that

of robotic rovers for future Mars missions. Most

A Small Enough Team

to

Do the Job

93

deep space missions were science-driven: Their design was molded to isfy

the requirements of the relevant

NASA-selected Principal Investigators. that

would

on

ride

it,

was

would not be permitted would have

a

mandate

community of

scientists

MESUR Pathfinder,

sat-

and the

and the rover

a "technology mission": Science requirements to drive

up

cost.

to cut capability

On

the contrary, Pathfinder

necessary to keep costs within

if

the specified budget.

The

rover's mission started out looking like a near duplicate of the

Mars Science Microrover demonstration. Once rover

would deploy

rock, chip

away

off the lander, the flight

a micro-seismometer, navigate across the terrain to a

rock (we hoped more successfully than Rocky

at the

and then, using an Alpha Proton X-ray Spectrometer instead of Rocky visible point spectrometer,

the way, the rover

after a

Donna

determine the composition of the rock. Along

MESUR Pathfinder lander, which might show

rough landing. And just by commanding the lander to

image the rover during the

Mars

properties of the

4's

would take pictures. Some of those pictures would doc-

ument the condition of the some dents

4!),

rover's travels,

surface,

and

we would

learn about the soil

how well the rover performed.

Shirley shrewdly declared that the rover mission

would be

only seven Martian days in length. Promising such a short rover lifetime

had two major payoffs. The

overall design of the rover could be simplified,

which would help keep the cost down. And

since the rover

had

a

good

chance of surviving beyond seven days, the probability of mission success

was high. The rover team could declare success

early

and then keep going.

Each additional day of rover operations would be another day exceeding expectations, pure gravy.

w-

The

first

sanity check took place in

Team Review"

In

new

empowered

project. Their job

team could get from here

was

was

called the

to there for the

to cast a cold eye

to assess

money

The Cognizant Engineers each presented design.

It

"Red

NASA parlance a "red team" was a group of technically

astute but disinterested engineers early stages of a

October 1992.

on the

whether the rover

available.

their pieces of the rover

SOJOURNER

94

Bill

Layman

cle design

described the eight- wheeled "rocker-double-bogie" vehi-

he was

partial to.

The

eight- wheeler

was an even more capable

would work equally well

variant of Bidder's rocker-bogie, and

driving

forward or backward over obstacles. While Layman currently preferred this design,

between

The

he made

and

four-, six-,

rover

it

team was

clear that the

considering tradeoffs

still

eight- wheeled concepts.

power source would

consist of a solar array

and non-

rechargeable batteries. Rechargeable batteries would have been eighteen

times heavier for the same storage capacity!

would not operate below

freezing,

And

rechargeable batteries

making them much

less resilient to

temperature fluctuations than the other electronics in the rover. For a short mission, nonrechargeables

would support

all

necessary nighttime

operations and emergency power needs while driving, while continuing to function at the lowest temperatures.

Layman

"The rover couldn't possibly work!" He wanted reliability

a

was probably

figured the reviewers' going-in assumption

them of the Layman pointed out

to convince

of the rover's design. So in his presentation,

He had insisted on putting enough

key feature of the power subsystem.

batteries into the rover so that even if the solar array failed completely ter landing, the rover

could survive and operate for a

hausting

And

batteries its

its

energy.

full

week before

by some catastrophe, the

if instead,

were drained on the way to Mars, the rover could

still

af-

ex-

rover's

complete

mission objectives using solar power only. The rover might be a low-

cost system, but his

team was thinking like

The red team listened to tions

a spacecraft design team.

the presentations for a

full day. It

asked ques-

and wrote recommendations.

The reviewers

did not believe the design the rover

team showed them

could be done for the money. But they did believe that there was a rover that

was doable

for the available

budget and schedule,

limited the requirements they were

must be

brutally prioritized."

The red team

"Go with what you know how ready existed: Use the rover

it.

trying to satisfy.

to

do

right

if

the rover

team

"The requirements

instructed Shirley's group:

now."

The Rocky

4 design

al-

Forget the eight-wheeler. Forget future options for

CPU: Commit

to the 80C85.

uncertainty: Maintain large

The

rover effort was fraught with

monetary reserves

—perhaps

as

much

as 50

A Small Enough Team to Do the Job

percent

—to manage the resultant

risk. The red team The reviewers preferred

Shirley's seven-day mission.

didn't

want the rover

erations also

95

to

as

Rocky

The message was:

The

rover

team

objectives.

To

across the terrain,

the red team,

we were

still

much with too little.

We all hated these interruptions to the "real work"

of engineering the rover. But the reality was that early in the project often put a spotlight

of reviews

this series

on issues that begged for solution,

but which the rover team did not have the authority to resolve on

one

That

listened. Mostly.

Other reviews followed.

In

They

4.

Simplify the design.

trying to accomplish too

thirty days.

same mission

to complete the

meant the rover could be dumber in its navigation

dumb

Donna

do more; they just wanted to give the future op-

team more time

almost as

didn't like

case, the Pathfinder science integration

plans for the rover, loading

it

down with

its

own.

team had grandiose

over nine pounds of science

in-

struments plus the weight of the devices that would deploy those instru-

ments from the

rover.

included not only the

Their proposed science payload for the rover

APXS and a micro-seismometer, but a neutron spec-

trometer intended to search for water. For a rover whose

total

mass could

never be greater than about twenty-five pounds, these instruments would

be quite a

ball

and chain. Worse, the

details

been defined. The science guys were stall its

design effort until they

views, the rover

of the instruments hadn't yet

implicitly telling the rover

made up

team

At one of the

their minds.

to re-

team presented the instrument mass requirement they

thought the rover could handle: 2.6 pounds. The science representatives objected and presented their

own number. The

review board declared the

science requirement unacceptable. Within a week, the proposed micro-

seismometer was gone. The neutron spectrometer was gone.

All that re-

mained was the APXS. The design could move forward.

While the rover had been funded for

tended to

fly

on had not yet received

some

final

time, the mission

it

was

in-

approval from Congress as part

SOJOURNER

96

of the

NASA budget. Once

nearly assured of

full

And then Mars

that

was taken care

of,

Pathfinder

would be

funding to completion. Observer, one of the most visible JPL missions,

launched months before, suddenly disappeared.

It

was loaded with instru-

ments, including a camera that would take images of the Martian surface

with better than

six-foot resolution.

just about ready to

sent the

The $500

million spacecraft had been

be nudged into orbit around Mars. JPL operators had

commands

to prepare the spacecraft. Part of the

command

se-

quence involved shutting down the communications transmitter, then turning

it

on

again.

But

after the

For days, attempts were

The antennas of any

signals It

made

NASAs Deep

sequence began, there was only

to send

new commands

Space Network (DSN)

to the spacecraft.

listened in vain for

Mars Observer might be sending.

seemed

that

NASAs

string of failures that

began with the Chal-

lenger disaster had not ended. This one pointed straight at JPL.

Congress view vote

this as a sign

down JPUs newest deep

For days

I

was

Mars Observer block.

silence.

of incompetence, a lack of worthiness, and space mission?

agitated as

in the

Would

I

contemplated

news and

MESUR

my

uncertain future, with

Pathfinder

Would I soon be looking for work? In the end,

on the chopping

despite the

Mars Ob-

server mystery, Congress approved MESUR Pathfinder. Those of us working the mission and its rover payload collectively relaxed. Now all we had to

do was get to Mars.

EIGHT

THE ROVER WAR

late

August 1992,

InProject

Donna

Shirley

and

Manager Tony Spear and the

on the current

Bill

Layman

briefed Pathfinder

MESUR Pathfinder project

status of the microrover design.

At

this point,

staff

only a

couple of months after the effort had begun, they had only a "conceptual design" to present. There were sign options.

ered rover.

One

A

still

many tradeoffs to be made among de-

of those trades was that between a tethered or unteth-

tether

would provide

a physical connection

between the

lander and the rover: Across that connection could flow power, nications,

An

and control

signals for the individual devices

untethered rover would have to provide for

power

source,

commu-

onboard the

itself,

carrying

rover.

its

own

—and through the lander

communicating with the lander

with Earth-based operators

—via

a radio link,

and relying on

its

own com-

puter for navigating over the alien terrain.

Spear favored the tethered approach; he said he feared that a radio link

between the lander and rover might be link at

on

a

moving rover

to carrying a portable

home, and wondered

if

the connection

one engineer on Pathfinder put

mean,

—

I

don't

it:

He compared the

phone from room

would be

to

radio

room

as easily broken.

"Tony mistrusted the

He had had some know what that was

really mistrusted radios.

perience in the past

unreliable.

As

radios. Period.

I

obviously very bad ex-

—and he saw the radio

SOJOURNER

98

money.

link as nothing but a rat hole for

rover

He also saw a situation where the

would be puttering along and you'd have

problem

a line-of-sight

with a rock between the rover and the lander."

Some would be

of the engineers in the meeting thought that a tethered rover a simpler rover,

with a direct "telephone

and not just because the radio could be replaced line" to the lander.

One of

the big problems for

warm in the cold Martian environment. (It was much easier to keep a large object warm at night than a small one. For such a small rover was keeping

its size,

a small object will have

a larger version of

itself.)

more

Electricity

rover across the tether, supplying

surface area exposed to the cold than

from the lander could flow

power

into the

to heaters that could maintain the

And with the lander acting as power source, the rover would no longer need its own batteries. Perhaps the rover could rover's

temperature overnight.

be simplified even further: eliminate the rover's computer, and let the

computer send

der's

signals directly to

lan-

each motor on the rover.

But there were problems with the tethered approach. The rover couldn't drag a hundred yards'

worth of cable around with

The

would catch on

would get

tether

break. Instead,

tangled,

you would need

line as the rover

drove along.

a tether spool

If

tangled in the ever-lengthening

The bottom ity to

line

perform

its

Layman and

was

that a tether

as

rocks, abrade,

it

moved:

and

finally

on the rover to play out the

the spool got line,

it

jammed, or

the rover got

then the rover's mission was over.

would

severely restrict the rover's abil-

primary function: exploring the surface.

recommended the untethered rover concept. Most of those in the room agreed. But Tony Spear was adamant about the tether.

The

Shirley

between

friction

beyond the

issue at hand.

the briefing ended,

it

Shirley

and Spear was obvious.

The two of them

seemed

It

did not get along.

that Spear didn't

clearly

By

went

the time

want any kind of rover on

his spacecraft.

When

Shirley brought the tethered option to the rover team,

Stone groaned.

He had just come

had been adapting

a

off of the

Hazbot

project, in

Henry

which he

commercial mobile robot platform to operate

in haz-

ardous environments. That robot had had a tether. Stone had personal experience with taste in his

how

mouth.

difficult

it

was

to

"manage"

He wanted no part of it.

a tether.

It

had

left a

bad

The Rover War

The concept of removing did not go over well.

of the lander,

the rover's computer from the design also

Not only would it make

was not

it

practicable.

Sending

erate the rover's actuators along the tether

tether

the

was

same

a cable as thick as

The

tronics to be almost as

the rover a

complex

the signals needed to op-

all

signals

on the rover

"separating out"

mere appendage

would not be

your thumb, the

wires, then be separated out

appropriate devices.

99

would

as the currently

easy.

Unless the

would have side,

to share

and sent to the

require sufficient elec-

planned rover computer

would be. All

could think of was

I

trailing

behind

it.

If it

was

how we would operate the rover with a tether a rule that the rover could never drive over

its

wrapped around an axle), then how we hope to drive up to a rock, take measurements, and then back away from it? One of the important characteristics of the rover was that it

own

cable (or risk getting the tether

could

could turn in place, so

hanging out ity.

rover

capabil-

an operations nightmare.

team knew in its gut

thought Spear was

rover

could maneuver in tight spots. With a tether

we'd never be able to take advantage of that

A tether would be The

We

in back,

it

crazy.

We

that a tether set

was the wrong way to go.

out to prove that the untethered

would work.

Layman viewed the tether fight as a wasted, lost year. How much more could we have done if the rover team had not had to expend its resources defending design?

but instead had devoted

Much later, Henry

he considered ing

itself,

this battle

now had

a

time to improving the

Stone would have a more positive perspective:

with Spear to be key in taking the people work-

on the rover and melding them together

personnel

its

common

enemy.

We

into a true team.

The

rover

would work together or

having the dream of getting to Mars perish in

its

risk

infancy.

A few months after the conclusion of the Mars Science Microrover demonstration,

Office

Tony Spear had hired Lonne Lane

as the Pathfinder Science

Manager /Rover Manager. Lane would be responsible

for coordi-

nating the development of the Pathfinder science payload: the lander

camera, the APXS, the neutron spectrometer, and any other instruments

SOJOURNER

100

might be

that

selected.

would manage the

And both

flight rover

development

turned from a Pathfinder-related that his

Donna

Shirley

Spear and Lane assumed that Lane activity.

trip to Russia,

had assumed the

he was amazed to discover

wanted

also

never received indication of an ad for such a position.

know why. The answer is: if it

I

know why

really

wanted to

is,

if it

was

a

competed

position,

The Pathfinder

project received

organization, while the rover

Code

Due

R.

I

know about it."

Pathfinder and the rover were funded out of separate pots of funds.

I

wasn't in the country.

those kinds of things. So, the answer sure didn't

He

to

Donna told me later it was And you just don t do

never saw the ad.

I

was,

re-

Rover Manager. "Tony had

role of

view of what was to be done. And Tony

publicly out. But

But when Lane

its

funds from the

NASA

NASA Code

S

was supported out of the research-oriented

to this organizational detail, the rover

was not

truly part of

the Pathfinder project, and Spear did not have any authority over the rover funds.

He had hoped

that the joint

appointment of Lane would put the

rover and instruments together under Pathfinder

ing coordinated development.

manage

the rover far

Now it was

management, ensur-

clear that Shirley intended to

more independently of

Pathfinder than Spear had

imagined.

Spear had a daunting job ahead of him:

He was

assigned the task of

doing a Mars landing mission for one-fifteenth the cost of the

last

attempt:

Adjusted for inflation, the Viking mission of the 1970s had cost about $3 billion.

Spear had only $171 million. Although JPL had accepted the chal-

lenge of the

"faster, better,

team was

the Pathfinder

cheaper" Pathfinder mission, the sense within

that JPL's upper-level

lieved the mission could not

reer and

were

was

management

be done. Spear was nearing the end of

willing to take the risk. But he well

hope of succeeding,

to have a

Beyond

privately be-

it

would have

knew

that

if

his ca-

Pathfinder

to be an extremely well-

the natural friction that existed between

him and

focused

effort.

Shirley,

Spear worried that efforts to incorporate the separately funded

and managed

flight rover

would

divert energies

away from the design of

was the heart of the Pathfinder mission. He could ill afford such distractions. And who knew if Shirley's rover would do what the the lander, which

mission needed?

The Rover War

So

in the

101

middle of July, Spear went to Lane and

his

Bob Wilson. As Lane remembered the meeting, Spear cost estimate for a derivative of the science rover that I

want to know what's the

system engineer said,

you

"Give

did, the

me

a

demo.

do one or two of 'em." Wilson and Lane

cost to

complied, putting together an estimate for two rovers that would deploy

APXS

an

spectrometer to figure out what rocks were

carry cameras, and might carry one

Their

came

estimate

first

Code R funds. There was

a

more instrument

in too high, if they

made

of,

would

as well.

wanted

to capture the

hard limit on the cost of the rover: $25 million.

Spear came back to Lane and asked, "Look, what's another approach?

Be

creative.

What

can you do

and Wilson mulled

it

you want

if

And then Spear said:

"I

want

this in less

Lane nosed around NASA's Code

had gotten

what

is

than

funded by Code R.

When he got them on the phone,

"Would you consider other approaches beyond or

different

Lane

from

being pursued on the existing microrover?" Their answer was an-

you back in

a

few days

after

say, "I

don't

know yet.

We'll

we mature some ideas."

Lane and Wilson went to work on the rover

we

weeks."

R to find out if they would be open

other question: "Like what?" Lane could only call

six

Lane

"All right."

know during the Mars Science Microrover effort, which had

to

partially

asked,

Then they said,

He made a few phone calls to people he

to an alternative to Shirley's rover.

been

to get the science back?"

over for four or five days.

study.

"Given

six

weeks,

assembled a four-person team." They pulled in a couple of engineers

from two aerospace companies they had worked with recently on other projects.

By

early September, they

Being creative often Lane's it

would have

means eliminating

team started with

to be finished.

much

as

complexity as possible.

a small rocker-bogie concept,

to a four-wheeled design. "It

was projected

der was going to be quite benign,

really,"

that the area

Bob Wilson

then simplified

around the

lan-

said. "So, trades to

money by minimizing the capability of the rover were definitely made. And that's where the rocker-bogie assembly went away. It went to save

four wheels. lot

We

of money."

weren't going to do rock climbing, period. That saved a

SOJOURNER

102

Then Lane and

team looked

his

nection between lander and rover. "The tether idea

do things that were very liver

at a

power. That

very high

made

rate, if

do image processing. line to

clever. If

of a physical con-

at the possibility

you did

came

along.

a tethered system,

the thing a lot smaller.

You could

We could

you could de-

deliver data

back

you so chose." You could use the lander computer to If it

got cold at night, you just sent power

down

the

operate heaters inside the rover.

Lane's team sat

down with

engineers at Hughes Aircraft, which had

developed tethers for wire-guided missiles. The tether they proposed for

was

the four-wheeled rover

a direct derivative of this technology.

were ways of wrapping a tether around the line without jamming. As Wilson

a spool that

it

would reliably release

commented, "When you were out

of tether, you were out of mission." But thin that

There

who

cared?

The

tether

was so

could be miles long.

Unlike Shirley's team, Lane's group focused solely on satisfying the anticipated science needs of Pathfinder. a need.

We

saw

a science need.

much muscle

about bringing

as

table for a finite

amount of

we were

given

We

—to do

a

said, "I will create to satisfy

we were

thought

as possible

—

on

science job

being quite clever

scientific

And we thought

dollars.

good

Lane

muscle

—to

the

that

was the charter

this mission.

We wanted sci-

ence return. You land, in a limited radius, what can you do

that's really

meaningful?"

The four-wheeled rover wouldn't have

a

computer onboard, except

for those built into the individual instruments

would be on the

rover computing

was not

like a

Rocky, but

would carry The only

lander, using the

Pathfinder lander relied on to operate. Lane cept: "It

it

it

was

same computer

the

was proud of the new con-

able to get the instruments out,

and do something with them."

Soon

after the

study began, Lane realized he had a big problem: "Into the

third or fourth week,

Code R had been the

first

it

was very

clear that

Code R had no

investing in rover research for years,

time, funding a flight rover.

They wanted

interest in this."

and was now,

for

the rover to be the cul-

The Rover War

mination of

103

of that research, an exciting practical application of the

all

new technologies that had arisen from Code R funding. They made no secret of their intent for the flight rover to

be a technology demonstration,

not a science payload; engineers were already busy defining technology experiments to validate the rover's engineering performance once

on Mars. Bob Wilson described

it

was

this

way: "There was a

desire for an aggressive rover doing aggressive things.

Our approach was

driving around

The

to support the science:

science doesn't

it

need to do aggressive

This was the fundamental disconnect between the

Code R

things."

research

agenda and the objectives handed to Lane's study team by Tony Spear.

So midway through the rover

"We It

think

we

Lane went to Spear and told him,

have an interesting concept that probably would be viable.

would do the

that

study,

science.

It

would be an

active rover

you could photograph and keep track

capabilities. Yes, this

is

what experience we had with MSM. But one

iota to the line to support this.

Tony,

is

Code

S going to step

I

I

would have some unique

truly believe that, based

don't see

So the question

up on the

money just was not there. So ument it, and close it out. the

After the six-week

of. It

something we can do.

out on the surface,

on

Code R stepping up is,

if

you want

this,

science side to support this?" But

Spear told

Lane to

finish the study, doc-

study was finished, Wilson and Lane examined

some

smaller and simpler versions of the rover, but even this activity had trailed off to nothing

by mid-October. To Lane,

concept was dead. that there

"It

would be

weeks before

it

was

quiet.

a review of

hit." So, just a

presented to their

team review board

own

Then

it

all

seemed that the tethered-rover of a sudden

we were

notified

what we had done. That was about two

few weeks

after Shirley's rover

team had

red team, Lane's tethered rover would face a red

as well.

The red team

for Lane's rover

sought to com-

pare his concept to Shirley's "free ranging" design.

Some

of the engineers from Shirley's team were in the audience,

ther to present aspects of their

own

ei-

design or simply to watch. Here was

Lonne Lane, who had managed the Mars Science Microrover, which had

SOJOURNER

104

sold the idea of a flight microrover, presenting a competing rover design.

For those of us on Shirley's team, Lane was

now

the enemy.

I

sat there,

ready to pounce on any weakness in Lane's presentation.

Lane

didn't even call his concept a rover, but the

Deployment Mechanism. He described ments If

it

would

carry, the

IDM, or Instrument

features: the tether, the instru-

its

simple navigation concept.

the review board had been skeptical about getting Shirley's rover

out the door for

its

$25 million budget, they were

more

posal to build the Instrument

Deployment Mechanism

$17 million. They asked Lane

how much

his

leery of the profor the

promised

contingency he had included in

budget and schedule.

Lane told the review board that he was assuming able

team of engineers would be working

achieve his proposed schedule

on

that his extremely

more hours

sixty or

a

week to

time.

Everyone knew that the engineers on

flight projects typically

worked

long hours to meet schedule. But those extra hours were something a

manager held in reserve already dependent

for

unexpected problems.

on those long hours just

If

you

started a project

to handle the problems

you

could anticipate today, you'd burn out your team before the surprises even arrived.

They'd have nothing

view board wanted to see

left

more

to give

when

the crunch came.

The

re-

reserves in Lane's plan.

When Lane prepared to bring in one of his non-JPL team members to discuss detailed technical results of the IDM study, the review board chairman vetoed the presentation on the grounds that the review contained JPL discreet information, and

no outsiders would be permitted

in

the room. In spite of Lane's protestations, the chairman held firm. Lane

attempted to limp along without one of his key technical contributors present.

Representing Shirley's rover,

Don

Bickler presented the impressive

performance of the rocker-bogie, including higher than

IDM

its

own

wheels.

felt

it

this

overall merits of the effort

would

collect.

So

we

to

over rocks

know how

the

unprepared to respond: Prior to the

meeting, "we were led to believe that

the data

ability to drive

The review board wanted

compared. Bob Wilson

based on the

its

was going to be

—the

effort

a decision point

being the rover and

all

put together our presentation addressing

The Rover War

that."

Lane and Wilson were shocked by the review's focus on mobility

capability. "That's all

shook

105

his head. "It

was decided upon.

it

was

six

weeks of

very,

We

were ambushed." Wilson

very hard work, incredibly long

hours, totally wasted."

The engineers on

knew

nothing of the

Shirley's

team who

sat in the

audience

specific technical objectives that

sign of Lane's Instrument

Deployment Mechanism.

the review

at

had guided the

de-

We judged the IDM

by the same technical requirements we were designing our own rover

to

We felt a Mars rover should be designed for rough terrain, and that cutting corners to simplify the system was a mistake. We didn't know that there was no money in the Pathfinder budget to fund the IDM. We meet.

thought the project was trying to fund

future of our rover depended

nology an

on

had

fight over Pathfinder's rover

R was

this review.

largely

had already occurred

room: Code

own

rover, just so

development, and to cut us out.

total control over its

events that

its

far

We

capabilities they

didn't realize that the

been decided

it

could have

We thought that the in

our favor due to

beyond the walls of

funding our rover because

it

would

this

conference

deliver the tech-

had always supported; they would never pay

for

IDM that served only the needs of science.

The IDM might be dead, but Spear link.

He

still

feared a failure of the rover radio

Donna

continued to pressure

Shirley to put a tether onto her

rover.

In January 1993, Carnegie-Melon University dition to Antarctica.

Under

NASA

funding,

mounted

a robotic expe-

Red Whittaker and

his

team

had built Dante, an eight-legged walking and rappelling robot. Dante was intended to descend into

its

own

bus active volcano. With great

"hell," actually the crater effort, the

CMU

of the Mt. Ere-

group had transported

themselves and Dante from Pittsburgh to Antarctica, to the slopes of Mt. Erebus, and then to the

lip

of the volcano. They had hooked up the tether

upon which Dante depended

for

power and communication. But

after the

The expedition was

over.

At the next rover team meeting, Donna Shirley conveyed Spear's

reac-

robot took only a few steps,

tion to the

its

tether snapped.

news of Dante's broken

tether: "Shit!

We'd better go

wireless!"

SOJOURNER

106

Two months trip to

later,

Pathfinder engineers and scientists

made

a field

study the geology of Death Valley and the Mojave Desert for

sights that

might impact landing and roving. During the

rover telecommunications to a range of three

in-

field trip, the

team demonstrated the proposed rover radios

hundred yards,

several times farther than

anyone imag-

ined driving the rover out from the lander on Mars. At the conclusion of the radio

test,

Tony Spear gushed his

relief,

congratulating the rover com-

munications Cognizant Engineer for demonstrating that the radios would indeed work better than his cordless phone.

NINE

DESIGN THAT REALLY WORKS

A

1994 the pace of the rover design

MFEX,

accelerating. its

short for Microrover Flight Experiment.

heavy reviews and threats to the existence of Shirley

was

had been christened with

flight rover effort

By

effort

official

name:

The period of

MFEX had passed. Donna

was protecting the team from undue reporting requirements by

declaring to the line

management that she would be the

sole source of in-

formation about team progress: Leave the rover team alone and

do

The

let

them

their jobs! It

built

universal.

barreling tions,

up slowly

No

one could say exactly when the feeling became

But there came

down the

track.

a

when the MFEX team felt like a train many months of frustration, distrac-

time

From

the

and apparent lack of progress, the team had emerged

leaner, deter-

mined, and with the framework of the design in place.

There were

lots

of engineering problems to solve, but the team was

handling most of them as

fast as

they were coming up.

A rover engineer

could look around and see the other engineers on the team pushing ahead.

The momentum of

the

team drove the

individual

members into

a

run just to keep pace.

At the weekly core team meetings, the Cognizant Engineer from

SOJOURNER

108

each subsystem would report the significant

activities in his

or her area

week. Often a key issue would come up, one that affected

for the prior

more than one subsystem;

the meeting could instantly segue from status

Many

reporting to a detailed design session.

sticky

problems were solved

then and there, allowing the design effort to proceed. The meetings

of-

ten ran over three hours, until the exhausted engineers dragged themselves off to lunch, only to continue the discussion as they ate.

During

the meetings Shirley constantly typed notes into her laptop computer,

recording everything. Within a day she would send out an email of

"Rover Significant Events" providing minutes of the meeting in excruciating

detail.

The Pathfinder project and its to influence the design

science payload people often attempted

—the characteristics of

and features of the rover

the rover's cameras, which instruments the rover

would

carry,

what

ex-

periments the rover's software would support. Sometimes these requests

would team.

not to

travel

When

Shirley,

informed

at the

but directly to the engineers on the rover

weekly meetings of such requests, Shirley

would remind her team of what she is

called

Donna's

not a requirement until someone pays for

desires than

money

to pay for them.

And

if

it."

'

rule:

Pathfinder had far

more

they weren't willing to pay,

Donna's rule was

Shirley did not feel obliged to be responsive.

warning to the rover team to avoid expending effort to quests that, taken one at a time,

A requirement

also a

satisfy technical re-

seemed eminently reasonable, but

to-

gether would ruin the effectiveness of the team.

The evolution of

the

MFEX

system was following a pattern of major

design problems uncovered and

new

challenges met. Each subsystem

doom the rover. The solutions to these difficulties would often pose new challenges to another sub-

would

deal with issues that could potentially

system. Bill

Layman kept the team moving.

those technical problems ria

—which

is

tightening the schedule screws

You

was important to manage

so that people didn't just dither

a natural reaction

nical requirements.

"I felt it

finally

of a committed engineer

on him, and not

relieving

—go into hysteif

you just keep

him of any tech-

reach a point where they really can't do

it.

A Design That Really Works

And

they

109

know that. And they don t know how to

ing logical.

The

proceed.

people are just nuts, setting these schedule constraints. at the

beginning of

if

you tighten the screws

everyone will eventually become to leap to the end.

the job done."

They

Layman watched for

of that that's a payoff: their intuition

is

Til just start

here

If

you

tight

enough, almost

because they try to figure

inefficient,

stop being methodical about getting signs of overload within the team.

But there was also an upside to the

and

stop be-

problem and work in an orderly way from the be-

this

ginning to the end/ But

some way

They

best engineers have the quality of saying, 'Well, these

stress.

"There's a certain

amount

force people to take bigger intuitive leaps,

more and more efficient as some optimum point where they

sound, then you can get

you force larger and larger leaps. There's

succeed with ninety percent of their leaps. For the ten percent they don't

make,

there's

time to remake that leap or go a more meticulous route to

the solution of that particular problem."

Layman

The optimum

laughed. "That's right. There's the

where you're taking

risks intentionally, to

optimum

level

of

stress?

level

of

stress,

accumulate a budget of time

and money, which you then spend to solve the problems where you missed your guess or the risk was too

large.

ating at just about that optimal level of risk. tent in every area

.

.

.

and overworked

I

felt like

The

our team was oper-

team was compe-

rover

in every area."

Layman had to deal with the dynamics between rover team members as well. One day the Pathfinder flight system manager dropped by his office and asked Layman how his new job as rover Chief Engineer was going.

Layman's answer: "The hardest part of the job

Eisen and

Henry Stone from

is

keeping

The

tearing each other apart!

rest

Howard of

it is

piece of cake." Eisen and Stone were the Cognizant Engineers for the

a

two

biggest subsystems, and they were constantly battling over technical ap-

proaches and

who had

agreed to what. (Surprisingly to

two of them partnered lieved stress

many

others, the

for sailing races after hours. Sailing together re-

and reminded them they were teammates.) The assembled

team was

full

of strong personalities. They had different

priorities,

and

distinct

ways of doing

tant characteristics, they

things.

skills,

different

But in the two most impor-

were the same: They were

all

good

at

what they

SOJOURNER

110

and they

did,

two

these

wanted

all

traits

to put a rover

on Mars. Those who did not share

on the

did not last long

project.

Layman's imprint was on every aspect of the rover design.

It

was there

in

how he got the most out of every member of the rover team. If an engineer came to Layman with a solution to the time," the Chief Engineer the problem," overwhelming bility left.

be

killed,

Layman was

a design issue that

would

it

often ask about possible ways to

so that there

insightful

risk.

an easy way out, Layman might ask him, before a review board to explain

them?" Layman called

Layman

it

why

"designing

was no

"kill

credible failure possi-

to know which problems had to And when an engineer proposed

enough

and which could afford the

would work "most of

"If,

years

from now, you stood up

the rover failed,

what would you

on the path of least

tell

regret."

continued to be concerned over power on the rover: "Every

spacecraft IVe ever

worked on had

enough power." You

at least

two

dealt with this sort of

crises

where there was not

problem by

either getting

more power to begin with, or using what you got more efficiently Usually both. The rover's primary source of electrical power would be its solar array, a flat

teries

panel that would cover the top of the vehicle. The onboard bat-

were there mostly

for emergencies.

Layman wanted

the solar array

to get bigger.

Solar panels are fragile things.

The mechanical team wanted

to keep

the panel out of the way, inside the footprint of the rover's wheels, so that a passing rock wouldn't shatter

any

cells.

With

a smaller panel, they also

wouldn't have to worry about collisions between the panel and the rover's

own

wheels, going

up and down on the

rocker-bogies.

The

Engineer pushed on the mobility team to make the panel sible,

rover Chief

as large as pos-

extending the array over the wheels: "The solar panel needed to be

as big as

we dared make it."

In the final

compromise, the mechanical team

cut out the front corners of the panel above the front wheels, just to keep the rover trol

from injuring

itself as

and navigation guys had to make

good enough

it

drove over rocks.

the con-

sure their navigation system

to keep the overhanging solar array

hazards of Mars.

And

enough was

away from the natural

A Design That Really Works

111

Because the rover would have to get by on the limited power available

from

solar panel,

its

only to the devices

was imperative

it it

needed

cluding a series of CPU-activated boards.

tem

When Layman

saw the

power switches

power

This necessitated

in-

in the rover electronics

inefficiencies in the

power switching

sys-

proposed by the electronics guys, he counterproposed: "Let's do

first

everything with mechanical relays."

was

that the rover brain supply

at a particular time.

possible, but

He

wasn't even sure his suggestion

he wanted to galvanize the team into coming up with a

better design. Perhaps partly because the electronics engineers found Lay-

man's mechanical solution so repulsive, they went away and then came

back with a

Layman

far

more

efficient solid-state design.

defined the state of rover thermal control as

it

existed at the start

MFEX: "Rocky 4 had a sheet metal frame and a pile of circuit boards on top of it. Nobody had a vision about how that could be configured to survive the Mars environment. It was clear immediately to everyone that we needed what amounted to a beer cooler that we put all the sensitive stuff of

inside of

and kept warm." The thermos bottle concept that would keep

the rover electronics tronics Box,

WEB

from freezing was quickly named the

for short.

The

WEB

Warm

Elec-

would warm up during the day

while the electronics were on and generating lots of heat. Then, over-

thermos bottle would slowly cool off. By the time it dropped the electronics' lower limit of -40 °F, it would be morning, the

night, the

close to

would wake up, and heating would begin again. Layman presented yet another ugly solution to motivate the

rover

the

WEB. He

suggested a big box to hold the electronics, with

sulation to keep

between the

Howard

WEB

warm. The box was

rover's

wheels and was

main

to

likely to

technical challenges.

it

bump

They

lots

of

in-

would hang down low into lots of rocks.

WEB set

as

one of

his

me-

out to improve Lay-

concept into something practical, something that would

work without degrading needed

so big that

Eisen recognized the design of the

chanical team's

man's

it

design of

make

the rocker-bogie's mobility performance.

the box smaller.

thermal protection of the

To do

that,

They

everything that required the

WEB —electronics boards,

navigation sensors,

SOJOURNER

112

radio,

and batteries

—would have to be packed closer together. Layman

took the position that the mechanical guys would never get

all

box they were imagining. "Show me," he

tronics into the tiny

the elecsaid.

And

eventually they did.

The thermal lem:

The

analysis of the

inside of the

matter how

WEB was

WEB

design revealed yet another prob-

No

going to get too cold overnight.

still

much insulation you layered on, the inside of the WEB would

eventually reach the average temperature of the Martian environment in

which

on

all

it sat.

run heaters plied

The

rover needed another heat source, one that

would

stay

the time. But they couldn't afford to exhaust the rover's battery to

The only option remaining was one commonly

at night.

on deep space

missions, but increasingly out of favor: Radioisotope

Heater Units, or RHUs. encased in a

C

ap-

An RHU

consisted of a tiny plug of plutonium

cell-sized graphite container.

about a watt of heat. The beauty of

it

was

A single RHU generated just

that

it

would continue pump-

ing out that watt for years, with only the slightest degradation. There

come anywhere close to this capacity tiny mass and volume of an RHU. But RHUs were controversial.

wasn't a battery on Earth that could

with the

Although millions of

dollars

to establish the safe design

had gone into detailed

studies over the years

and handling of RHUs, the public remained

fearful of anything radioactive.

Eisen set about determining the feasibility and cost of putting three

RHUs into the MFEX rover. He soon learned two facts. First, there were a number of RHUs already in existence, spares manufactured for the Galileo spacecraft, but never used. These RHUs were in the custodianship of the Department of Energy. Second, there was no mechanism for charging a all

new

project for the use of those

of the appropriate procedures, the Eisen's

team

also

had

to find a

RHUs.

So,

once

RHUs would be

good way

MFEX

completed

free!

to insulate the

WEB. The

standard methods of insulating equipment in space depended on the

presence of

vacuum

side the walls of the

—they counted on nothing being

landing, the tenuous atmosphere walls, creating a

there.

Vacuum

in-

WEB would work fine on the way to Mars. But after would seep

"thermal short,"

like

an

into the space

between the

electrical short circuit, that

would allow heat to flow too freely out of the WEB. The only way to keep

A Design That

Works

Really

113

die Martian atmosphere out would be to put something else the

WEB

in.

At

first

engineer settled on powdered aerogel, a lightweight material

that was an extremely good insulator. You could pour the powder into fiberglass

honeycomb

that

WEB walls.

formed the

aerogel arrived from the manufacturer, ise

vi.

it

was

The rover team had been desperately

But

when the

a third heavier than

trying

stay

::

prom-

within the mass

Xow we were in trouble.

allocation.

Within weeks, Daw: riuun. an engineer on Eisen versation with Peter Tsou. Tsou

s

ream, had a con-

was a JPL engineer with

a facility that

could manufacture small quantities of an alternative form o: aert gel

Braun thought

solid form.

sulating the at

there

WEB.

would be

this notion: if

them

into the

Greg Hickey suggested redesigning the

honeycomb

material at

tion, similar to that

all.

He proposed

you

in-

tried

honeycomb.

Those gaps would form more thermal

air gaps.

—

might do the job of

that solid aerogel just

But there were problems with

small pieces of solid aerogel and insert

third engineer

use

the

batch of

shorrs.

A

WEB wall—don't

a sheet-and-spar construc-

of an airplane wing. There would be space for aerogel

between the inner and outer

WEB

where

walls, except

fiberglass bulk-

heads would link the walls at regular intervals. The solid aerogel had to be

made in slabs of precise thickness (the aerogel would crumble if you tried to machine it). To ensure that there were no air gaps in the walls. Hickey would build the walls around the slabs of aerogel, with the walls pressed tightly against the aerogel,

compressing it.

Hickey assembled a sample

When "solid I

; ;

air."

I

held a piece of aerogel in

The

piece

uldn't feel the

was

weight of

it.

I

Using

on

I

was prompted

a side

and yet so

to call

a smokiness in the aerogel his

own name

One

it

light that

looked through the material to see

my

of the en-

for solid aerogel:

heaven."

solid aerogel, the

mass of the insulation

dropped by over two pounds. For

pounds

worked!

my palm.

on the mechanical team had

"manna from

It

a couple of inches

palm only slightly obscured by gineers

WEB wall

total, this

weigh:

_

a rover that is

inside the

massed

tremendous

all

WEB walls

of twenty-two

SOJOURNER

114

Sometimes the most mundane elements of the design would prove to be the very ones that required the greatest ingenuity to be

One such

area

was the

motors on the rover was

on in the

among

all

of the boards, sensors, and

collectively referred to as the "cable harness."

MFEX design effort, there was no one responsible for the ca-

ble harness. But as the design

had come

to work.

rover's wiring.

All of the connecting wiring

Early

made

began to mature, Layman knew that the time

to find an engineer to oversee the rover's wiring. After a brief

Layman brought Allen Sirota onboard the rover system team. The last time I'd seen Sirota, he'd been leaving the Robby team to

search,

turn to his

shaken bring luck,

Sirota's

him

love

first

—

a flight project.

hand and

back.

Now

told

him

shook

I

his

At the time, a few years

that if

hand

we

re-

earlier, I'd

ever built a Mars rover, we'd

again.

Through coincidence and

we had made good on that promise.

Soon diagram

Sirota

was the master of the rover interconnect diagram. This

didn't

show

the details of the computer boards; instead,

showed how those boards would connect Wires led to every device. The

APXS

on the

rover.

electronics, the rate sensor, the ac-

celerometers

—

cable started

on the boards and split into

all

to everything else

it

would need power and data lines running to them. One several connectors that

mated

to

connectors on the internal bulkhead; from here, cables ran back to the other devices inside the tronics boards snaked

WEB. Another

cable emanating

that

out in the cold Martian breeze. As the design of the gressed, the wiring diagram

them

at

went through many

would be

MFEX

revisions,

sitting

rover pro-

each more

last.

You couldn't just dering

elec-

through a tunnel to the outside of the rover to

power and control the motors, cameras, and sensors

elaborate than the

from the

string wires

from one component to the

next, sol-

each end. Engineering must go into ensuring the rover

could not just be put together, but taken apart as well. Once assembled, the rover fail

would be subjected

and require

repair,

to tests, then inspected.

Components might

or entire assemblies might be swapped out. Over

time, subtle design flaws might be discovered, requiring modifications to

the electronics boards. So removable connectors were good:

They

al-

115

A Design That Really Works

lowed the design to be more modular. But they

up

added mass and took

space.

The number of wires was the design stabilized, there ics

also

boards out to the sensors and motors on the rover. So

would form

a thick bundle, as thick as

handle;

difficult to

it

would be too

you placed the bundle down there

was

was

to

bend very much, so every time and

installed the boards,

cold. All

a

to repair

it,

chance you'd break other wires in the process.

a

would be even

wires

your thumb. This would be very

inside the rover

And suppose you managed

rover

stiff

electron-

many

good chance you'd break a wire. Then you'd have

a

and there was

Mars was

By the time

starting to look like a problem.

would be 243 wires leading from the

to install

all

that wiring without

damage.

those wires leading outside into the frigid Martian

stiffer

complex

air

and more subject to breakage than on Earth. The

set

of moving parts.

Many

of the wires would need

How long would they last?

to flex as the rover traversed across the surface.

Worse, the copper that made up the wires was not only a conduit for

power and was

data.

It

also

conducted heat. The purpose of the rover's

to keep the electronics

and sensors inside

warm

WEB

despite the extreme

many wires routed from inWEB's warmth would be leaking out

cold of the Martian environment. But with so side to outside,

much

of the

through them.

The mechanical team's

solution to the heat leak

tunnel, often called the "igloo tunnel"

nel

was

a mini-labyrinth

on the

was

to create a cable

front of the

WEB. The

tun-

through which the wires would be routed. The

convoluted path the wires would follow meant that the cables would be

much longer,

several feet,

more slowly along good

solution, but

the labyrinth.

and the warmth of the

WEB would flow much

The igloo tunnel was a make hairpin turns inside

the wires into the deep cold. it

How

also required the cables to

would the

cable bundle reliably survive these tight

bends?

Something

The

in the design

issue stayed

on the

was going

table for

to have to change.

many months.

Eventually,

John Car-

done, the designer doing the mechanical layout of the cabling, offered up a suggestion.

He recommended

using "flex-cable" technology, which

SOJOURNER

116

would eliminate

A ribbon would bend easily in some directions, almost not at

ribbon cable. all

So long

in others.

in use,

of the separate wires, integrating them into a single

all

as

you anticipated the

you could design

a flex-cable that

without stressing the "wires" inside

ers.

flex

could be bent almost in half

Flex-cables

it.

had to

were manufactured like

with alternating layers of conductive "traces" and insula-

circuit boards, tion.

directions the cable

A flex-cable for the rover might require twenty-five or more such lay-

Cardone had designed layouts of

He'd

flex-cables for other projects.

never seen one as involved as what the rover would need, but there was no reason, in principle, that Sirota

it

couldn't be done.

and the rover team studied the

idea.

JPL did not have the

build flex-cables in-house, so going with flex-cables

ties to

facili-

would mean

And a complex manufacturing process would cost money. But in this case the money seemed like a good tradeoff against the contracting

them

out.

development schedule

risk to the rover's

liable design. After evaluation

cable approach

had

Sirota

Sirota

was

if

they went for a cheaper, unre-

and debate within the rover team, the

just hired Art

Thompson onto

and Thompson had worked together on

Space Shuttle

the

MFEX

system team.

Sirota's last assignment, a

experiment. Sirota immediately assigned

flight

to be the contract

manager

for the flex-cable

Thompson

procurement.

Pioneer Circuits got the contract. There would be three types of cables,

one

inside the

The

flex-

WEB between the two electronics boards, one from

the boards to the internal bulkhead, and one that tunnel.

flex-

in.

went through the

igloo

tiny rover's igloo tunnel flex-cable, as designed by Sirota and

out by Cardone, would be the longest and most complex flex-cable

Pi-

oneer had ever produced. From one ribbon coming off the boards, the

ca-

laid

ble

would

split into six

connector.

The

units, spares,

sand

dollars.

among the

of flex-cables would be expensive: Between the

sets

and But

ribbon "fingers," each to be soldered to a separate

test runs, the contract it

should be worth

was worth

it,

several

flight

hundred thou-

creating a high-reliability link

rover's electrical assemblies.

Not everything went according to

When the first set of ful inspection.

The

cables

plan.

began

a care-

had been delivered without connectors;

Sirota

flight cables

were

delivered, Sirota

11?

A Design That Really Works

had separately purchased flight-qualified connectors, which would later be affixed to the flex-cables

As part of

by experienced JPL

needed to confirm that the proper "wires"

his inspection, Sirota

in each finger of the cable aligned

with the appropriate solder location on

He placed the first connector against one of the

the back of the connector. fingers of the flex-cable.

flight electronics technicians.

It

didn't match. Sirota quickly

checked the other

connectors and fingers against each other. They didn't match Sirota

went pale. Something was fundamentally wrong. The arrange-

ment of the wires image of what

it

in

each finger of the flex-cable was backwards, a mirror

should be! They had just spent $450,000 on the contract,

with most of that pulled out of reserves. It

was

how

instantly clear to Sirota

ger of the flex-cable had been connector, called a "micro-D."

D

And the

cable

was

useless!

the error had occurred. Each

fin-

made to be soldered to a specific type of They were called D connectors because,

when you looked at them face-on, ter D.

either.

they had the shape of a

tall,

connectors had two genders, male and female, that

skinny

fit

let-

together.

Male connectors possessed pins that plugged into the female connectors' Micro-Ds had been chosen to minimize weight and volume:

sockets.

They were the

smallest and lightest D-type connectors available. Their

small size forced the pins to be recessed, so they looked a lot like sockets.

And the

sockets looked like pins.

carefully

It

was easy

to get

them mixed up.

Sirota

checked over the JPL drawings. The error was there, a reverse

terpretation of his intent. Sirota

blamed

himself:

He

in-

hadn't personally

checked the drawings that had already been reviewed by two other engineers. Pioneer Circuits

had simply complied with the

specifications in the

JPL drawings and designed for male micro-Ds instead of female micro-Ds. Other than

was

perfect.

this

one huge

The obvious choice would be would

basically

mean

the rover schedule] sible."

do? to completely redo the design,

starting over.

was so exact

There was no time

left

implementation of the flex-cables

error, the

What was he going to

to

".

at that

do the

.

.

and the timing. The timing [of

point that

realizing that the flex-harness

thing else

would have

would just be imposThose were

"What helped me

would have

to be reworked."

it

flex-cables over again.

the things bouncing around in Sirota's head.

was

which

eventually

to stay the same. Every-

How hard was that going to be? "I

SOJOURNER

118

had the

realization that

most of the

stuff hadn't

been wired

stuff that

had been wired could be reversed. Then

could

through

live

"Basically, still

felt

I

a

yet,

little

and the

better.

We

this.

we reworked nine cables here, and told everybody who was

going to supply their cables to reverse them, because they hadn't

We caught it really quickly. It could have been a lot worse.

wired them yet.

That was something maybe not many people knew about.

been

a disaster, but that

was

Sirota

was averted very quickly by some

number of

silent for a

seconds.

". .

.

It

could have

fast thinking."

That's called dodging a

bullet."

Small though the rover was, cles before

it

was

still

too big. Like

this feature also

to safely traverse over rocks higher than

made

the rover

the rover and between

find a rise

way

up

to

on the to

tall

its

to

it

wheels. But

fit

inside the

And all that empty space under

Somehow, the mechanical team would have

down

the rover crouch

to

while inside Pathfinder, then

selected

Ken Jewett

to be the mechanical engineer

responsible for the overall configuration of the rover. So figuring out

make

it

height once on Mars.

Howard Eisen had to

high

wheels would have been wasted volume while

lander.

make

its full

its

was too

taller. It

Pathfinder lander during the trip to Mars.

sitting

of the Rocky vehi-

the flight rover's rocker-bogie running gear gave

it,

ground clearance

was

all

the rover stowable

fell

how

onto Jewett 's shoulders. Where Eisen was

loud and combative, Jewett was quiet and

self-effacing.

He just wanted to

do the design work, solving the fundamental questions

that stood in the

way of making a mechanical system function: "The most creative part is in the design. The rest can get excruciatingly boring." Jewett didn't usually make a lot of noise, but he did have a temper. He didn't like it when other people tried to

make

laziness or lack of

their technical

from the

would need

side,

his

problem, due to

good engineering skill.

Stowing the rover was see that he

problems into

his

problem. Jewett struggled with

to "break" the rocker. If

the rocker

was

the bigger of the

you looked

two

it.

He

could

at the rover

pieces of the rocker-

A Design That Really Works

The rocker had

bogie.

a

wheel

at the rear end.

1 19

and came forward

pivoted freely around the forward end of the rocker. The rocker

named

oted around the arbitrarily the rovers

the height of the rover

WEB

was

touched the deck

When

until the

the time came,

two

the rover to stand up. and those

permanently into

place,

the rover's mission

;-.

would be

a low-rider

feated by the smallest

:

all

the

reduce

nearly

some mechanism would pieces

force

would return

to

would then have

to

pieces of the rocker

with no chance of slipping back out. Other-

would be

with so

:

WEB

bottom of the

The rocker

their original "unbroken'' shape.

latch

The only way

side-to-side.

break the rocker into two pieces. Then the

to

down

rovers body could drop

itself piv-

an axle that went

"jeff tube."'

from

it

on each end.

attached to the bogie at the front: the bogie, with a wheel

way through

until

over: at

its

stowed height, the rover

ground clearance

little

that

it

would be de-

rock

How should the critical latch be designed? Je we tt studied the problem, finding

no easy answers. He

left

the latch problem alone for a while and

worked his other rover design assignments: deploying the mounting the cameras get in the

so that other

way and generally just coming up with clever ways

rover lighter. But the latch

That was

a

was always waiting

hard nut to crack.

where there were linkages

signs

that

would drop

and

slots there

into a [notch]

was

I

kept putting it

that

came

make

:

the

off.

We tried several de-

together, pins running in slots it."

But with pins

something would bind, or a pin

:he chance right.

a sufficiently reliable version

:

for him.

and couldn't get out of

would get bent and stop working would be

rover's antenna,

components of the rover wouldn't

Jewett just didn't trust that there

of these approaches that met the

rover constraints. "I

fooled around with that design for about a year before

how

an Aha!' and figured out

was

to use a particular spring

two pieces of the folding it

jus

its

I

that

anchored

really

work." That inspiration

in just the right spot to

rocker, it's just a bent spring

into a locked position.

locked position, yet

would

it's

And

it's

final>.

I

very strong

and

very flexible before that." The

simpler than the earlier alternatives he had examined.

as

when

each of the

it

comes

it's

final

up.

finally in

design was

Simple

is

good."

SOJOURNER

120

Unlike the other designs,

this

one

"just

wanted to work." Once he had the

design concept down, he could stand back and appreciate

something apart from himself,

made: "Maaaaan,

as if

elegance as

that's neat!"

Layman pushed on the mechanical team

way of mounting

its

were something discovered and not

it

again, this time to find a secure

on the lander

the rover

Not only

deck.

did the

body of

the rover need to be held in place during launch and landing, but each of

down separately as well. The team worked out a way to tie the rover down at three points, so there was no possible way the vehicle could accidentally come loose. Cable-cutter pyrotechnics, fired by the lander after it was safely down on the Martian surface, would the six wheels had to be tied

hooks held each wheel

release the rover. "Cowcatcher" cleats.

Only the turning of the

them from

rover's

in place

by

its

wheels under power would release

their restraints.

How was the rover going to get stood up so that the springs on either side of the rover

would get the chance to snap? One of the

early "standup"

concepts was to put a big spring under the rover. As soon as the tie-downs

holding the rover to the petal were released, the spring would push on the

bottom of the WEB,

lifting

Layman

the rover upward.

did a calculation

showing that any such spring would be so powerful that the rover would literally leap into

gravity.

Not

Eisen

a

the

good

air,

"as high as

your shoulder,"

in the

low Martian

idea.

came back with

a

new

idea: Let the rover stand

up 'by

itself. If

the rear wheels drove forward, and the rest of the wheels stayed in their places, the rockers

would

eventually the rockers

was

simplicity

itself,

start to

would

lock.

elegant in that

lift.

Keep driving those wheels, and

To the mechanical team, it

involved no additional hardware or

mass. But to the control and navigation team, careful testing.

this solution

it

meant new software and

How would the rover tell that it was done standing up? If

the wheels drove for too long, their cleats

would

break, and the rocker-

bogies would be overstressed. There would be potentiometers on the bogies to

measure

how the

bogies

moved during

traverses

(How big

a

rock

A Design That

did to

we just drive

over?);

maybe

Works

Really

121

the readings off the pots could be adapted

monitor standup.

Henry Stone wanted contact would be

that

triggered

when

on the

sensors

the rockers locked.

give a sure, positive indication that the vehicle ters

would have

results.

to be carefully calibrated,

The switches would

had stood

up. Potentiome-

and might provide

less certain

Eisen complained that contact switches were not particularly

added more mass to the

able,

rockers, small switches

reli-

and required more wires going in and

rover,

WEB, which would contribute to the loss of precious heat from inside the rover. He proposed that the bogie pots would be good enough, out of the

and they would be there anyway. Eisen convinced Layman that the rover could do without contact sensors. Stone wasn't happy, but accepted the decision.

The

lack of those contact switches

headaches, and Stone would

come

would eventually cause many

to regret not having fought harder to

keep them in the design.

As the rover's design matured toward completion, events conspired to take

our key leaders away from ble.

The

project asked for

us.

The

Pathfinder lander design was in trou-

Layman s

help to get the lander out of the dol-

drums. The lander team was uncovering issues faster than with them. The lander was overweight, exceeding

There wasn't enough room

for

all

could deal

mass

allocation.

of the subassemblies to

able volume. Certain issues with the

lander were

its

it

fit

fundamental structural design of the

And

not complete, and the clock was running.

still

in the avail-

clear that the landing cushion airbags

would

survive

it

wasn't

the lander

if

came

down with a significant horizontal velocity, which it could if the wind was blowing on landing day. Layman did not want to move off the microrover team, because he didn't

feel

seriously to leave the rover,

team

relied

on me,

that

I

he had finished the job.

and go try to

was trying

to

couldn't do that.

I

got pressured pretty

the lander.

I

felt

was an important part of the

kept the machinery turning in the team. that

fix

"I

abandon

lubrication that

My response to the management

move me wholly onto couldn't

that the rover

this

the lander

was

team and expect

that they just it

to function,

SOJOURNER

122

because

we had built a team that was one deep

'one-deep' person in

my particular spot,

system architect." Layman was reluctant to but in the end

der,

would never get

felt

he had

He

to Mars.

little

He knew

that he

I

shift his attention to

was the

the lan-

choice: If the lander failed, his rover

accepted the role of Pathfinder Project Me-

chanical Engineer as an additional duty piled neer.

everywhere, and

which was kind of the overview

would not be

on top of rover Chief Engi-

able to maintain a balance

between

rover and lander.

The

rover design activity

was winding down, moving on

and

test.

The lander was not

as far along.

to assembly

He was

sure to be

sucked in by the lander, "inevitably drawn further and further away from the day-to-day doings of the rover."

Layman began

passing pieces of his rover responsibilities to Sirota,

Stone, Eisen, chief rover system engineer Jake Matijevic, and me.

hoped

that the

team would have

really guilty, like cal

I

moment. But

tially

had

up

set

assuring that the lander

moving

chance to transition gracefully.

this organization,

could see no

I

a

way

would

to

then backed out

do anything

fail."

to the lander side of the house,

I

As

I

else

learned

found

it

were so many problems remaining on the lander

He

"I felt

at a criti-

without essen-

why Layman was

frightening that there at the

same time

that

we were finalizing the rover design. Would the lander ever come together in time?

And then,

Donna Shirley was promoted out of her job as Rover Manager. The number of Mars missions was growing. NASA wanted

a

in July 1995,

whole

series

of missions, done in the

"faster, better,

cheaper"

mold, with two spacecraft launches every two years. Responding to

new NASA mandate, JPL was

creating a

Mars Exploration Directorate.

Shirley

new

this

internal organization, the

was chosen

as

its

director.

At an all-hands rover team meeting, Shirley announced that while she

would

try to

be the

official

keep

someone with take

hand

a

in as

long

as possible,

Jake Matijevic would

Rover Manager. She then said that the team would need the

same thoroughness and attention

on many of his

duties.

to detail as Jake to

As we walked back from the meeting,

of eyes on me. Allen Sirota smiled and put a hand on obviously you," he grim, as

if

now

said.

I

wasn't so sure, but

my

I

felt lots

shoulder.

"It's

Henry Stone was looking

the decision had already been made.

"I

don't

know what

I'm

A Design That Really Works

going to do.

Where am going I

I'm screwed."

But

I

was

also

Within

a

I

was committed

committed

few weeks,

to find

someone

123

else to

to Stone's control

do what you do?

and navigation team.

to the rover as a whole.

I'd

job for Stone and half of a

been

split in

new job

two, trying to do half of

for Matijevic.

my

old

TEN

THREE ROVERS

and navigation work on the

control

The Building

107: software

navigation algorithm testing.

first

had heckled

Henry Stone had moved his

at a talk

I

office into

MFEX effort.

encounter with Stone had been a few years

me

earlier

when he

gave on Semi- Autonomous Navigation of the

Robby rover. He had asked question after question about a detail of the gorithm. At rail

first

I

had wondered

my presentation.

detail-oriented

in

development, sensor design, electronics and

the building almost immediately after joining the

My

was centered

flight rover

al-

he was maliciously attempting to de-

if

later realized that

Stone was simply both incredibly

and unusually persistent.

He had an inherent need to delve

I

into every aspect of a technical concept,

understood every one of those

details.

and

This

didn't feel satisfied until

trait

he

contributed greatly to his

success as an engineer, although he sometimes forgot that others in the

room were not

necessarily vitally interested in such a complete analysis.

Stone seemed to be happy only fast pace.

deal,

things

were moving along

at a

The other engineers he worked with, who respected him a great

sometimes kidded him about

was absentmindedly pressing seemed he was trying centrate.

when

his

to crush his

his quirky habits.

hands against

own

head.

It

his

One

of those habits

temples so hard

seemed

to help

him

it

con-

125

Three Rovers

Henry Stone smiled wistfully as he considered the

MFEX:

"I

actually

first

naively thought that we'd have such a small

be involved

tem job became grew, Stone

in the software design."

clear,

and

team

that

I

would

As the scope of the subsys-

his responsibilities as

was forced to let go of this

few months of

Cognizant Engineer

desire. Instead,

he was pulled more

and more toward the most complex hardware that would come out of the control and navigation subsystem: the electronics boards comprising the

custom

rover's

brain.

These boards would control everything the rover

operating every device onboard.

did,

Gary Bolotin joined the team

as the lead rover electronics engineer.

His job was to design the rover's computer.

The meat of

new components, but

not be the creation of

his

work would

the selection and arrange-

ment of particular combinations of existing components

into circuits that

did specific useful things. There were circuits to switch the rover's motors

on and

off,

and reverse

ues to determine

up the rover

their direction; electronics to read out sensor val-

how far the

at the

proper time after

and on and on. The

teries;

rover had driven; an "alarm clock" to it

had shut down

result of Bolotin's

to conserve

work would be

wake

its

bat-

schematics,

virtually a paper representation of the thousand electronic components

and

of the interconnections

all

among them

—that would together

consti-

tute the brain of the flight rover.

The custom hardware ality

that

would embody that design and make

re-

were the printed wiring boards. Within those boards would be

etched

all

of the necessary

circuitry.

every component would then be place.

The volume

Once

the boards were manufactured,

mounted and soldered

constraints of the rover's

The components would end up

onto the boards that

would be

The

it

flight rover couldn't

available commercially.

Not

a

into

its

reserved

WEB forced a hard limit on

the size of the boards.

bility.

it

wonder if they

so tightly packed

all fit.

use just any electronic parts that might be

all

parts

were

built to the

same

level

of

There were standard commercial components, and "military

fication" parts designed to reliable

of

all

were Class S

work over wider temperature flight-qualified parts,

relia-

speci-

extremes. Most

intended to survive harsh

SOJOURNER

126

space environments. sion,

If a particular

type of part failed during a space mis-

an advisory would be issued warning of the

that part.

There were engineers

such suspect parts found their

at

JPL whose job was

way into new

was the most a

his design, parts that

detailed spreadsheet that

I

to

make

with

sure

no

spacecraft.

Bolotin put together a spreadsheet to track

had incorporated into

risks associated

all

of the components he

needed

ever did

—

to

be purchased.

for anything."

"It

He gave

copy to Stone. The next time Bolotin saw the spreadsheet he had

thought was complete, Stone had taken

was tracking the

and enlarged

it

it

enormously.

He

status of every part destined for the electronics boards.

For some unusual parts, small enough to

fit

on the crowded

electronics

boards, the estimated delivery dates were twenty-two weeks after receipt

of the order. That was a five-month lead time!

How do you develop software "Rapid prototyping" thing

is

when you have

the process of building a practice version of some-

barely the information

vealing problems in this

Rocky

would

be:

for a flight rover that doesn't yet exist?

way was

the

first

you need

big step to solving them.

4's

wheelbase was almost the same

As

a rapid prototype,

team gutted Rocky

4,

it

would

removing the

to get started. Re-

do.

size as the flight rover's

Howard

Eisen

mobility

s

electronics, instruments,

and rock

chipper used for the Mars Science Microrover demonstration. They

re-

placed the existing wheels with wider-track stainless steel wheels, added

prototype steering mechanisms and sensors, and then they drove totype rover through

of Martian

—

soil

soils

fine

this pro-

with a consistency that matched our best guess

and powdery

like

talcum powder. Once the me-

chanical engineers had satisfied themselves that the rover just fine in alien soil, they turned

their prototype vehicle

would perform

back over to the

control and navigation team.

Rocky 4 now became

a dedicated testbed for exercising the software.

Stone emphasized the point by calling the vehicle the

Development Model. His team commercial 80C85

CPU

SDM—Software

installed a small cardcage containing a

and wirewrapped electronics boards that

gether duplicated perhaps half of the functions of the future

to-

MFEX brain.

12?

Three Rovers

Accelerometers and a rate sensor measured the

and an early version of the the

summer

was,

it

rover's radio received

of 1993, a year after

version of the

SDM

was

MFEX

and turns of the

and transmitted

rover,

data.

had gotten under way, the

operational. Limited

would allow the software guys

though

to try out their

this first

By

first

testbed

motor control

algo-

around the building. Most of the time they

rithms, driving the rover

would just keep the

tilt

vehicle

up on

they could test their ability to

a stand,

command

its

wheels hanging in the

air,

so

each of the motors without the

rover wandering off anywhere.

team was going

The control and navigation

to be driving the

SDM vehicle

around. They would need a test area insensitive to the vagaries of the weather, one that looked to the rover like the natural terrain

someday

navigate.

Henry

building. Carpenters

cleared out

hammered

it

would

most of the biggest room

together a

in the

wooden frame about

eight

inches high and fifteen feet wide by thirty feet long. Forklifts drove

through the roll-up door into the frame.

Soon

it

When the

was time

at the front

dust settled, the rover sandbox

to put a

more

hardware that reflected Bolotin

Model"

—

a printed

computer to boards: the

of the building and

s

realistic

flight

dumped sand

was

ready.

rover brain onboard.

The

design was to be the "Engineering

wiring board version of the rover brain. Like the

follow, the

first

flight

Engineering Model actually consisted of two

CPU board containing the 80C85, memory, and most of input/

output circuitry; and the power board, mostly containing the converters

and regulators that powered the

many onboard

Within JPL was a group devoted to side Building 103

devices.

flight electronics fabrication. In-

were cleanrooms and "flow benches" designed to

vent any particles in the technicians soldered

air

from

settling

on

electronics boards as the flight

down components and

wires. Their procedures

been worked out over the years to maximize the tronics, since,

reliability

of

had

flight elec-

once launched, repair was impossible. The rover control

and navigation team contracted with 103 to turn Gary Bolotin ics

pre-

into the Engineering

The weekly

reports

Model

s

schemat-

boards.

by the Building 103 engineer coordinating the

SOJOURNER

128

electronics fabrication for a particular

seemed

week

into a disquieting pattern.

fell

weren't done, but we

The

to say every week.

times. Stone took

him

still

The

tasks

promised

have plenty of time, he

fabrication engineer said

it

one too many

off the team.

The engineer was supposed to be ensuring that all of the steps of fabrication were moving along rapidly to guarantee on-time delivery of the seem to understand an important rule of flight

product. But he didn't

final

projects:

A day lost now is a day lost forever. Launch dates don't wait. You

don't have plenty of time.

Stone took

He and

tion.

track.

It

upon himself

it

Bolotin

to bird-dog the electronics board fabrica-

would together

would take us months

on

see that the fabrication got back

to recover

from the delays introduced dur-

ing that dismissed engineer's tenure on the job.

Stone thought all

it

would be

good idea

a

countdown clock to remind

to get a

of us located in Building 107 of the deadline

He went within JPL. everyone.

to

George Alahuzos

He had been

to get

it.

we were working toward.

Alahuzos was an institution

there since the early

Stone sometimes called him the

sixties,

Sgt. Bilko

and he knew

of the Laboratory. To

me he was expediter par excellence. Over the years, Alahuzos his

henchmen, one or two

was teamed with Jim system, or a that

it

Lloyd. If

way around

got done.

If

the machine shop

skilled technicians, to

Now he

you needed something pushed through the

bureaucratic roadblocks, Alahuzos

you needed

fast,

always had

do the footwork.

would

to get a piece of flight hardware

made

.

.

.

We

were

just glad

he was on our

side.

It'll

electrical

be beautiful," said Alahuzos. Time went engineer to do some

that the clock

might show up

work

Whenever he questioned Alahuzos, still

working on

it.

More time

"I

Alahuzos found an

of months, but

the answer

passed.

We

would launch before the clock was ready

of

know just what you

by.

in his off hours. Stone

in a couple

how

So Stone gave Alahuzos a

small budget to find, procure, or build the clock.

want.

in

Alahuzos could get your work order slipped to the

top of the queue. You just didn't want to dig too deeply into the it

see

it

had thought

didn't happen.

was the same: They were

started to joke that Pathfinder

to

announce

it.

*

129

Three Rovers

When

Stone

he had commissioned

was

saw the nearly finished

finally

a rectangular

he was both pleased and appalled.

it,

box four

inch-tall digital display

clock, almost a year after

feet

wide and almost two

The upper

TO LAUNCH." The

would dominate any wall

clock

who

designed

it

had placed

programmed

box was

it

It

Alahuzos's technician

was

on the

a data port

on.

into

The clock was

mounted

left

to

do and

how

It

the

was

countdown clock on the wall

a constant

reminder of

time remained to do

little

A

it.

I

how much couple of

moved our

of-

the second floor of Building 230, the Spaceflight Operations Fa-

fices into

team.

lit

side so

computer hooked

via a laptop

months before launch, Brian Cooper, Henry Stone, and

cility,

a sign,

also reeked of "boondoggle."

overlooking the rover sandbox. there

six-

was mounted

the Internet to get an accurate time synchronization signal. beautiful.

It

"MARS PATHFINDER TIME

tubes, that read

the engineer

high with a

feet

half of the

from behind by fluorescent

that the time could be

was huge!

along the bottom marking the remaining days,

hours, minutes, and seconds.

And

It

joining the rest of

We

what was becoming the Pathfinder operations

brought the clock with

above the cubicles that covered the

us,

mounting

it

high up on the wall

The clock was big enough

floor.

to

be

seen and read from anywhere on the Pathfinder-owned section of the sec-

ond

floor.

After launch, the clock would be to

reprogrammed and the

sign

changed

"MARS PATHFINDER TIME TO LANDING."

The installation of components on the Engineering

Model

electronics

boards had just not happened as quickly as needed.

When

Stone went

over to Building 103, he had often found the electronics technicians

signed to his task off working ally

doing the work, the

came

up,

not

flight technicians

were

When

excellent.

they were actu-

But if

a

question

to a halt.

the supervisors in 103 were telling Stone that his schedule

realistic.

to find a

else's job.

and they needed to consult with Stone or Bolotin, the board

work would come

And

someone

as-

Stone could not accept

wav

to

meet them.

It

this.

He had

was obvious

deliverables,

that there

was

was

and he had

a lot

of dead

SOJOURNER

130

time on his boards. Only job,

and

infeasible.

if

the technicians were working full-time

The problem was

his

group was

that the electronics fabrication

service organization, supporting ticular

on

behind, would he believe that his planned schedule was

still fell

many projects

a

There was no par-

at JPL.

reason for them to consider Stone's job more important than any

other customer's. Every job for them was a flight job; why should they be more committed to Mars than to, say, Saturn? The experience of building the Engineering Model boards convinced

Stone that he needed to find a better way, or he would never succeed in delivering the flight boards

on

time.

To Stone, the

He needed

assembly people part of his team.

solution

was

to

Building 107, right next to the control and navigation team's

cated electronics technician. That

from Stone.

If

would

also put

dedi-

just downstairs

them into 107 would protect them from the

and ensure that they were working on only one

tractions in 103,

rover electronics boards. Stone began pushing

chain to try and to be

them

own

the

into

they had questions, there would be somebody right there

to answer them. Bringing

wanted

make

them moved

to get

make

it

up the

line

dis-

task, the

management

happen. The section manager responsible for 103

accommodating, but he

people, even temporarily. Stone also

resisted giving

went

up control of

his

own section manager.

to his

One day at lunch Stone announced that there was going to be a meeting between him and the two section managers about how to deal with the slipping schedule pleased, he

on the

Rather than being

was worried. Stone never expected much help from line man-

agement. For whatever reason, that intervention

He

electronics boards.

by the

on Pathfinder was

tended to slow things down.

line organizations

figured that the managers

other's territory,

his experience so far

would want

to avoid stepping into each

which was exactly what Stone's proposal

called for.

He

expected to get "eaten alive" in the meeting.

Stone was

at a crossroads,

ing the luxury of sitting

on the outside of the

pieces of advice. "Whatever

meeting.

And

if

you go

and he wasn't sure

you

in there

Bill

handle

situation,

do, don't raise

without

how to

I

it.

offered

Hav-

two

your voice during the

Layman, you're

crazy."

I

knew he needed someone with sufficient stature at the Lab to back him up. Layman was the only person I could think of. Henry protested that the

131

Three Rovers

meeting was only a few hours away, and he didn't know

At the meeting

in the 103 section

sues, laid out his case,

and argued

bly activity into Building 107.

would not

for

The

manager's

moving the rover

easygoing as you could imagine.

crowd was, and how a

breakdown

specifically the

listed his

The room was

is-

electronics assem-

solutions

tense. After the dis-

Layman had

his say, as

He knew how good a group the 103 He also knew how

excellent their products were.

team Henry was

table as a

Stone

manager suggested

section

require such a move.

office,

cussion had gone around in circles a few times,

good

Layman was

"Then get on the phone now"

available.

that

if

in

leading.

Layman viewed

communication.

It

the problems

didn't really matter

on the where

problem came from; the solution was to get the

right

people talking to each other. Locating them together in the same place

would make

that a

whole

The

lot easier.

Pathfinder mission and

were hugely important to the Laboratory, and we it

took to make

ger than

had

to

turf.

a success. This

it

was bigger than

all

had

to

its

rover

do whatever

section boundaries, big-

He was sure that everyone in the room could agree on what

be done for the future of JPL.

The

deal

working in

was

struck. Stone

his

two

flight technicians

107.

He had the people. He technicians

would have

would have the

still

needed

tools to

the 107 electronics lab space.

do

a facility in 107 their

work.

It

where those

was time

flight

upgrade

to

And Stone knew how to get it done. He put

George Alahuzos on the job. Flow benches, exhaust hoods, and door seals appeared. In a matter of a few weeks, yet another part of the building had

been transformed. Building 107 was ready

The

first

rover that

MFEX would build from

Integration Model, or SIM.

every

way

The SIM

to the final flight rover.

vehicle

The

rover

sembly procedures and refine them from gether the SIM.

If

to assemble flight electronics.

scratch

would be

identical in almost

team would develop

their as-

their experience putting to-

there were mistakes made, they

would be made on

SIM, early enough to avoid repeating them on the

would do much more than

would be the System

flight unit.

validate assembly procedures.

the

The SIM

During

its

voy-

SOJOURNER

132

age in space and

its

mission on the Martian surface, the

flight rover

would

encounter extreme conditions of temperature, acceleration, and vibration.

The SIM would be

ronmental qualification

a guinea pig, forced to tests,

undergo

a series of envi-

experiencing conditions far worse than any

we anticipated the flight rover would ever need to endure. The rover team was now engineering a design intended to survive these environments. If the SIM passed through the gauntlet of environmental tests, we would know that the design was correct. But the qualification tests would be har-

—the SIM would be aged by them. We would never

rowing

flight rover

When

before

its

risk

aging the

voyage had yet begun.

SIM would become

the time came, the

the team's "hangar

queen," substituting for the flight rover in operations tests as

hearsed maneuvers her

Second and

last off

Flight Unit Rover, or

FUR, we would

sister

the

FUR.

would perform

carefully apply

The FUR would never

roll

that its

it

would come the

of the lessons learned from the SIM. soil,

was Mars. The

or even be exposed to the

FUR would be

tested just

to prove that every aspect functioned properly, but not so

much

would experience any appreciable wear to its components or loss to

remaining useful

Donna

line

eventually built and handled the

through sandy

outside world, until that world

enough

all

re-

millions of miles away.

handmade assembly

When we

we

Shirley had

lifetime.

It

would be

pristine.

decreed that the rover's gender was female. Together

with the Planetary Society, a space-exploration advocacy organization headquartered in Pasadena, Shirley also worked out a plan for naming the flight rover, a

plan that would involve the public in the mission.

would be named

after a heroine, real or fictional.

The

rover

Only students would be

given the chance to submit candidate names, along with essays describing the key traits of the potential rover namesakes, and

would help the

intrepid rover carry out her mission

In January 1995, the

"Name

how

those

traits

on Mars.

the Rover" contest

was announced

in a

magazine distributed to science teachers around the country. Teachers told their students,

and the

entries

began flowing

in.

By the

deadline,

133

Three Rovers

3,500 entries had arrived, not only

Canada, India,

Israel,

from the United

States,

but also from

Japan, Mexico, Poland, and Russia.

names and

Selecting the best

essays

now fell into the hands

of a small

group of volunteers consisting of members of the Pathfinder and rover teams, and Planetary Society entries, style

staff.

Each volunteer read seventy-five or so

house on Catalina Avenue that served

When identity.

I

heard the

name

The word meant

was going

freed slave

who

"Sojourner,"

"traveler,"

I

of the Planetary

thousands of

knew

possibilities.

the rover had found

which was exactly what the

to be. Sojourner Truth

had

its

flight

lived during the mid- 1800s, a

preached for the abolition of slavery and equal rights for

women. The SIM would take the name of the after the

of the craftsman-

as the offices

Society, to debate the appropriateness of the

rover

room

then joined with the others in the living

chemist

who had

second-place entry, Marie Curie,

discovered radium and polonium in the early

years of the twentieth century. Matt Wallace, one of the rover

system engineers, worried over the Curie, a brilliant scientist,

name Marie

Curie.

The

power sub-

original Marie

had died of radiation poisoning from the same

elements she was famous for discovering. Wallace feared that the

name

Marie Curie, so long associated with radiation, would draw attention to the Radioisotope Heater Units

vent the

— RHUs—carried by the

flight rover to pre-

WEB electronics from freezing while on Mars.

Nevertheless, both winners were 1995, the thirtieth anniversary of

announced

to that public

from completion. But both of them far,

14,

Mariner Four's flyby of Mars. The SIM

had become operational only weeks before. The

them no matter how

on July

how wide,

FUR was

still

months

now had names that would go how long they traveled.

or

with

ELEVEN

SEEING AND BELIEVING

As

members

of the

Robotic Vehicles group were being "deputized"

out of the research

camp

into the flight project

community, Brian

He

Wilcox picked up the reins of the rover research program. tended to make the research program

do

so,

one

Wilcox would need

available:

Rocky

had inherited the

The

3.

as relevant to

a microrover of his

MFEX as possible. To

own. Fortunately, there was

Dave Miller had departed JPL, and Wilcox's group

vehicle.

old Mars Science Microrover had successfully demonstrated

CARD navigation of a microrover.

Its

onboard hazard detection sensors,

however, had been virtually useless in sunlight. The

would

in-

new Mars

rover

require something better, a "look-ahead" sensor, able to identify

obstacles before the rover

bumped into them. Wilcox knew that the "ma-

chine vision" software his group had developed for the

not an option. The 80C85 microprocessor on the

Robby vehicle was

new flight rover was far

too slow to handle the necessary computing.

The

human

flight rover didn't

have to have

its

own

path planner onboard:

A

operator on the ground would look at stereo images from the

Pathfinder lander's camera and designate the rover's path, relying

on

the

same basic Computer- Aided Remote Driving technology Wilcox had first demonstrated almost a decade

earlier.

The

lander's

camera was

to be

135

Seeing and Believing

called

IMP: Imager

for

Mars Pathfinder. To the

project, the

IMP was

a sci-

ence payload, not an engineering system. Several science teams had pro-

posed competing designs for Pathfinder's imaging system. Most of the designs had been for one-eyed cameras, without stereo capability7 Fortu.

nately for the rover team, the project just the kind of

images

had chosen the IMP, which produced

CARD required to function. The IMP Principal In-

vestigator, responsible for the camera's ter

development and

delivery,

was

Pe-

Smith of the University of Arizona.

IMP camera, Earth-bound

Using Smith's

engineers would be able to

plan paths for the rover that avoided obvious hazards. But, as long experi-

ence had proven repeatedly, dead reckoning error would cause the rover to drift off course during cles its

traverses, bringing

its

human masters had carefully worked to

it

face-to-face

with obsta-

And there might be

avoid.

hazards that the Earth-based operators would miss, especially as the rover forayed farther and farther from the lander and the

To protect

itself

from unexpected hazards, the rover needed

Here was

look- ahead sensor.

IMP camera.

a well-defined

a reliable

need where the remaining rover

researchers could directly help out the flight project. Wilcox

had studied the

various navigation sensing approaches researchers had used over the years for robot navigation.

that

made

He began

to consider variations

sense for a microrover. Wilcox

with would have to

knew

on these approaches

that whatever he

satisfy the typical flight rover

came up

requirements: minimal

mass, low power, small volume, limited computation.

He

settled

on

a variant of

an approach that had been implemented by

both robotics researchers and manufacturing industries again and again: "structured light."

known tells

a

The

pattern of light

you

camera

a lot

idea behind structured light

down onto

a surface, the

is

that

if

you put

a

bending of that pattern

about the shape of the surface. With the right pattern, and

to see

rocks, drop-offs,

it,

the rover

would be

and steep slopes

able to determine the presence of

directly ahead.

Wilcox's scheme required cameras on the rover.

For a while,

it

looked

like the

rover

would have no camera

at

all. Bill

Lay-

man remembered: "We'd been wrestling our way through budget cutting

SOJOURNER

136

and schedule cutting and we reduced scope to get people to believe that I

we

drastically several times trying

could build our rover for reasonable

Donna was ready to abandon anything

think at that point

talked a lot during that period of time about

Layman

I

mean she

how the rover's just going to

—

bump into things figure out where it's going by man without a cane. "Everybody we talked to that could provide us a

have to run around and

bumping into

...

cost.

things," like a blind

continued.

camera wanted more weight, power, complexity, and schedule than the

whole rover had,

camera

just for the

been thrown down that we needed dreds of thousands of dollars,

if

a

itself."

Wilcox: "The gauntlet had

camera and that

not millions of

it

would

dollars, to

cost hun-

put a camera

on here, big and heavy and have its own box, and its own power supply,

own processor, and an interface

of some form." Wilcox saw that going to

the camera development experts at JPL

had no experience delivering

its

was not going

to work.

hardware for the pittance

flight

They just

MFEX had

available.

Without

would

a

camera on the

apart.

fall

And

rover, Wilcox's

having no

hazard detection concept

ability to detect

hazards could easily

—

prove disastrous for the mission. Wilcox had another idea his

own—but it was

for a

would be no support from

clear that there

camera of Shirley for

developing a camera: She had already been convinced any camera would

be too expensive. So Wilcox went to Layman. Layman

summarized

later

Wilcox's proposal and their subsequent discussions: "Brian Wilcox

forward and

said, 'Gee,

vision camera,

and we'll mode-switch, and

sion for a while.

The called a

shouldn't take too

It

it'll sit

Charge-Coupled Device, or

CCD. The

CCD

is

and think

like a televi-

many components.'"

light-sensitive sensor that "sees" the

contains a

came

why don't you just let the rover's brain be the tele-

CCD

image

for short.

is

an electronic chip

Every

digital

arranged as a large array of

camera

pixels, corre-

sponding to the pixels you would see on a television screen. The camera lens in front of the strike the surface

responding

pixel.

CCD focuses an image on its surface.

of the

The

faster electrons will

electrons

on the

CCD, and

cause electrons to build up in the cor-

brighter the light hitting a particular pixel, the

accumulate

CCD

Particles of light

form

at that location in the

CCD

array.

The

a pattern of electrical charges that corre-

Seeing and Believing

13?

sponds to the image focused on the face of the CCD. The heart of an electronic camera, but lenses that

it is still

CCD

is

the

only a part. In addition to the

must focus the image on the CCD, there

are also electronics to

read out the built-up charges associated with each pixel.

Once

all

the

charges have been read, the charge must be flushed out, leaving a clean slate,

ready for the next image to be recorded.

TV camera, this process of reading out and flushing the pixels must happen very rapidly, usually thirty times per second. A typical CCD chip might have about 350,000 pixels, so the other electronics that For an actual

make up the camera ends up reading out over 10 million pixel values each second. The rover's 80C85 microprocessor wasn't fast enough to do that. But

we

didn't

need

video" from the rover; there would never be

"live

enough communications from Mars tion. All

day.

we wanted from

And for once,

Wilcox explained

"My

them

would just

in slowly,

sit

Mars would

to control the

out there on the

and gather the images full

help.

CCD

was cold you could clock the images

out to be fifty-three seconds for a read out the

was

basic concept

it

informa-

was one or two images per Martian

the hostile, frigid environment of it,

from the CPU. Because the charge

the rover

much

to Earth to send so

As

directly

in slowly;

CCD and you would just clock

as fast as

you

could,

which turned

image." Using the rover's brain to

CCD would also be convenient for hazard detection. Wilcox's

approach to obstacle finding would require only a small part of an entire image, a few horizontal rows of pixels, also called "scanlines." Wilcox:

"When you wanted to get range data you would flush the the vertical transport registers, so

scanlines out of

you got one or more rows you wanted

to analyze, so

you shipped out only those rows. So

you could get

a

in a

few seconds

few selected rows out of the image and do processing

on them."

Cameras onboard the rover would have other uses was convinced they would be purposes.

A

rover imager

of selected targets

main forever

both

scientific

distant

as well.

Wilcox

and public

interest

would capture unique close-up photographs

—rock formations and Martian from the

progressed, the public travels.

critical for

lander's

would want

dirt

—that would

IMP camera. And

to see the rover's eye

re-

as the mission

view of

its

latest

SOJOURNER

138

Layman wanted

a

camera on the rover almost

okayed spending "a few tens of K"

camera

idea.

dollars

as

much as Wilcox. He

on investigating the rover-brain-

Wilcox drafted Jack Morrison,

a software

and electronics en-

know much about CCDs, yet: "Brian CCD, and minimal electronics, and in-

gineer in his group. Morrison didn't

had the concept of taking a bare

terfacing that to the rover computer.

how CCDs

So

work, and then designed a

type to interface to that

A/D

I

quickly learned that

little circuit

all

we

I

could on

could proto-

board, and got that to work. There was a

long period there that started out with just trying to get the thing to work,

and then getting better and better images erate

it

properly.

basis for

The

electronics that

what we put on the

About the same time Shirley asked Brian

all

the time, learning how to op-

we ended up

designing

became

the

rover."

that Morrison

was designing

his circuit,

Donna

Wilcox to meet with the JPL camera experts and

plain his rover-brain-camera. After

all,

there

was no reason

ex-

to ignore JPL's

existing experience base in flight

camera design. And Shirley knew

Wilcox was an expert

not in cameras; he might be able to use

some

in robotics,

pointers.

At the meeting, the

overall reaction to Wilcox's plan

work." The main objection was to the

was

many seconds it would

"It

won't

take for the

rover brain to read out an image. Wilcox's concept basically used the as a storage device for the image.

CCD,

the

more

visual noise

The longer

would creep

that

into

it,

image had looked

time, and Wilcox

like.

The amount of

was proposing

stray electrons building

to

noise

to read out the

CCD

image remained on the

up and washing out the image. After enough time had

would overwhelm the CCD, making it impossible inal

that

passed, the noise

know what the

orig-

was proportional

to

image hundreds of times

slower than was traditional.

But the noise was also proportional to temperature: the lower the temperature, the slower the buildup of noise. Wilcox was counting on this.

Perhaps the camera wouldn't work so well on Earth, but Mars would

be colder. Most of the time the Mars

air

temperature would be below

zero, so the impact of noise would be minimal. The others in the meeting

remained

skeptical.

What happened

next was vintage Wilcox. Rather than resort to fur-

Seeing and Believing

139

ther technical analysis to convince the skeptics, he proceeded to complete a

working demonstration system, and do and Morrison

tion of Wilcox's

s

efforts

Within a week of being told that age,"

Wilcox

ture,

you could

office,

said.

The

picture

was

came it

and

liquid nitrogen, ff

freezer.

We

cooled

it

and equipment

was on the other

else

down

until

first little

cold,

had nothing but

on the

lines

then pulled

camera was a bare

over

ice all

it

out just

in a pot-metal

CCD

and

a

few by-

inside of the box. Everything

The only problem

there

We had to put desiccant inside the box,

wouldn't get

it."

The

colder the

was

so that

ice

when

CCD got,

the

images got.

Wilcox gave "Bill

It

was very

it

Our

side of the cable.

lens.

clearer the

s

in the makeshift

we

it, it

Morrison

and stuck the prototype camera

buildup on the

cooled

in

cooler with dry

on the voltage

pass capacitors

room tempera-

at

filled a picnic

screwed on.

a lens

but even

And to prove that the camera would work at

sat.

long enough to take an image.

box with

The culmina-

quickly.

a bit noisy,

Mars temperatures, Wilcox and Morrison ice

a shoestring.

wouldn't work, "we had an im-

clearly see the tabletop

where the camera

on

it

a printout of the first picture to the rover

Layman posted

the

first

were barking up the wrong

image on tree

.

.

.

Chief Engineer.

his wall, to tell the naysayers they

We

were very pleased to have

Bill's

support."

They would still need to design flight lenses and custom camera housings to shrink the size

been proven. The

would not break

and mass. But the rover-brain-camera concept had

flight rover

Wilcox's structured light tian sun. if it

would have cameras. And those cameras

the bank.

system would need to be brighter than the Mar-

The pattern of light produced by

were

visible

as far

from the sun

tiny power-limited rover generate as

secret

would be

lasers.

ergy in the form of heat and

wavelengths of

would be

even against sunlit ground. But the sun

even on Mars, half again

The

the rover

as the Earth.

very bright,

How

could a

light as the sun?

The sun pours out huge amounts of

light.

visible, infrared,

much

is

useful only

But that

and

light

energy

ultraviolet light.

A

en-

spread over

all

laser channels

all

is

SOJOURNER

140

its

energy into one wavelength.

length of

light,

If

you

the laser light can be bright;

wavelength, the laser

Wilcox went looking

bright laser diodes.

can be

invisible. Lasers

is

side laser pointers. (In fact, for early

pointers.)

are looking for just that

He found

indoor

if

you look

any other

for

the type found in-

tiny, like

testing,

one wave-

Wilcox did use

laser

commercial supplier of small but

for a

one, and found the lasers he needed.

The

lasers

put out light only in a particular wavelength of infrared, making

them

invisible to the

human

eye,

but very apparent to the

CCDs

in his

new rover-brain-cameras. To make the cameras even more sensitive to the lasers' particular infrared

cameras' lenses,

wavelength, Wilcox installed

that blocked

filters

all

filters

over the

same

light except light the

"color"

as the lasers generated.

Wilcox drew up

a simple optical design to spread the single spot of

laser light out into a fan.

draw

a line of light

Once

constructed, the "stripe projector"

would

on the ground. Wilcox made assembling the hazard

detection system sound easy:

"We

got a camera running on the cardcage.

We put that on Rocky 3. We got some lasers, built up some optics, and by summer

the

of '93 had a complete system running with

stripers."

The

lasers, all

mounted together on

final

sion processing and

configuration had

CARD,

a rigid

five

two cameras along with the camera

careful alignment

bar. Just as

laser five

with stereo

vi-

and calibration would be

necessary to a successful system.

The autonomous

traverse capability of the rover

was

activated

by the

"GO TO WAYPOINT" command. The command told the rover the coordinates of

its

destination,

measured

in

meters from

its

Given these coordinates, the rover headed straight for the off only if

it

Once

target, veering

encountered obstacles along the way. After avoiding any such

hazards, the rover tion.

starting point.

would doggedly return

the rover's

own

to a path

aimed

estimate of position told

it

at its destina-

that

it

had come

within about four inches of the target, the navigation software declared victory and the vehicle stopped.

Making "GO detection sensor. first

turned on

lasers.

TO WAYPOINT" work To check the

its lasers,

territory

depended on Wilcox's hazard

ahead for

took pictures with

its

safe passage, the rover

cameras, then shut

down the

(Keeping the lasers on only as long as needed conserved power.)

If

Seeing and Believing

the

ground just ahead of Rocky

formed

The

3

was perfectly flat, the

symmetric crisscrossing pattern of straight

a

stripes

in a rover

together

five lasers

on the

lines

surface.

of laser light would always be visible in exactly the same place

camera image. But

if

the ground wasn't

deformed by the presence of

shift,

141

the rover's slow- thinking

but only four selected rows of age, a laser stripe

pixels.

would show up

row The amount

a rock or a ditch.

CPU, the brain

didn't

To

simplify the job for

examine the

Examining

would

the stripes

flat,

a single

entire image,

row from an im-

as a single spot, the brightest pixel in the

the spot shifted

left

or right along the

tional to the height of the rock or the

depth of the

row was proporBy putting

ditch.

to-

gether the results from five laser stripes and four image rows, the rover created a sketchy topographic

wheels.

The map was just

whether there was

checking while

would

sitting

drive forward a

The hazard vehicle so that

around

drive

The

flight

detailed

a hazard in

a bit to the right, the left,

map

of the terrain just in front of

enough

If

for the rover brain to identify

view or not, and whether that hazard was

or directly ahead.

still.

The

rover did

the path in front of

when

was

it

all

of

its

hazard

the rover

clear,

few inches, then stop and take another look.

detection system had to look out far

it.

its

a rock

Wilcox

enough ahead of the

was detected the rover could

didn't

want the rover

stop, turn,

to be required to

and

back up:

design couldn't afford the mass and complexity of putting a

second hazard detection system on the rear end. Driving blind backwards

would be dangerous. The only way place until there

ward

know

were no obstacles ahead of the

in the direction the vehicle it

was even

The

process?

safe to turn

was now

it

rover,

But

made

and then drive

hitting

something

The

were too

for-

in the

sides in front of

the rover a cautious near-

encountered two rocks, one on either

path along which the rover might pass, the vehicle would around.

to turn in

how could the rover

had to look out to the

the rover, not just straight ahead. This sighted creature. If

facing.

around without

laser stripe projectors

up was

to avoid backing

side, leaving a still

rover could only see about three feet ahead, and

closely spaced to allow the vehicle to turn

if

often go the rocks

around while between

them, there was a chance the path led into a box canyon that could be

es-

went looking

for

caped only by backing up. Rather than another way.

risk

it,

the rover

SOJOURNER

142

This necessary feature of the rover's navigation algorithm was often frustrating to observers, even those of us

was

acting in such a timid fashion.

viewpoint of a

human being

who knew exactly why the rover

From

the comparatively omniscient

standing nearby in the sandbox, the correct

path to the rover's destination was obvious. But the rover's point of view

was more

it

rolled

and we would not be human perspective while

that of an infant crawling along the floor,

able to provide the rover with the benefit of our

through the dust of Mars.

Wilcox and Morrison continued refining the hazard detection and navigation software, running Rocky 3 around in the Building 107 sandbox.

They rearranged their design that

and over

again.

a hazard,

how

course for easily

its

grow

too soon,

it

the rocks periodically to discover any weaknesses in

might be masked by operating

There were far

should

still

it

questions to answer.

drive

away from

destination? If the rover

into a long one. But

if

it

same

in the

terrain over

When the rover saw

until

it

swung too wide,

resumed

a direct

a short path could

the rover turned back toward the goal

might run into the original hazard again, and be forced to

waste time avoiding the same rock a second time.

And what if

the tallest

part of an obstacle happened to be between the points in the rover's topo-

graphic map, making strike a

it

invisible to the vehicle's sensors?

balance in the hazard detection software:

The

They needed

to

rover could be too

bold, always driving over traversable rocks, but sometimes failing to see a real obstacle until

its

bumpers ran into it or it got stuck; or the rover could

be timid, almost never running into an obstacle, running away from something

it

lasers

it

largely proven, the

as the baseline for the flight rover.

onto the

SDM

expense of often

could readily traverse.

With the hazard detection approach accepted

at the

vehicle as part of

its

They

installed

MFEX

team

cameras and

upgrade to Rocky

4.2. Fine-

tuning of the rover's navigation system would continue until nearly the

end of the

MFEX development effort. w-

There would be only one test of the flight rover's hazard-detection system

under Martian conditions

—

at least

before

it

reached Mars. For

this test,

Sojourner would join the Pathfinder lander inside JPL's twenty-five-foot-

143

Seeing and Believing

diameter solar /thermal /vacuum chamber, and practice

moves

its

in sim-

ulated Mars sunlight.

So that the team could observe the it

had been sealed shut,

the middle of the floor rover.

We

activities inside

we

it

of the chamber, stood on

its

all

we had worked so hard to keep clean.

Instead,

we

green.

it

preparations, the Pathfinder lander sat in the center petals

its

after

would have contaminated

had welded the "rock" out of sheet aluminum and painted At the end of

chamber

placed a rock to serve as an obstacle for the

couldn't use a natural rock, since

the flight hardware that

the

TV cameras had been placed at a few key spots. In

open

as if

had recently landed. The rover

it

designated petal, ready to drive

off.

The chamber door was

shut and sealed. Following a carefully defined timetable, the environment inside the

chamber slowly transformed from Earth

to the temperature

A huge lamp at the top of the chamber stood in for

and pressure of Mars. the Martian sun.

moment on this simulated first day on Mars, the commanded the rover to drive forward. The rover

At the appropriate rover test conductor

complied, stopping only after lander ramp.

commanded

Now it

then head on to

would

its

force the rover to encounter our carefully

as bright.

left,

new

sequence, correcting for the rover's

around

it,

of Mars sun-

shouldn't it

work

.

.

.

could have detected the

command sequence. He sent a new position and directing the

rover test conductor aborted the

its

original target.

We

the rover to move. Instead of

watched the video monitor, waiting

moving forward,

was something in its way. The

There was nothing nearby for

The

it

long before

The

ing?

circle

rover tests outdoors in Earth sunlight, which

There was no reason

rock.

there

would

it

as designed, in the brightness

rover veered off to the

vehicle to

we

floor,

original destination. This traverse should prove that

We had already done

The

was down on the chamber

When the rover saw the rock,

whole system functioned,

light.

had cleared the end of the

to activate the hazard detection system and go to a way-

constructed rock.

was twice

rear wheels

that the rover

point destination that

the

its

rover

seemed

it

floor

there.

around the rover was

its

own

for

started to turn, as

to see as an obstacle.

to be scared of

from hazards that weren't

it

totally

if

flat.

What was happen-

shadow.

It

was running

SOJOURNER

144

While the rover team scrambled to

figure out

what was going on, the

lander guys were incredulous. Couldn't the rover drive a couple of yards

without getting into trouble?

"You don t know where

The lander mands kept

test

it's

We could read the expressions on their faces:

going, do you?"

continued.

We

resorted to "low-level" motion com-

to direct the rover to a safe parking location.

eyes closed,

its

When

it

we

did everything

asked of

for another project's use. After a

floor.

filters

chamber just

seem unusual

on them, making them

range of infrared

had

a

few days

to be cleared out

a grid

When we pulled up all of the tape, and turned on the

rover navigated in the

tape didn't

it

we had laid down to form

Mars "sun" (and got out of the chamber

The

without complaint.

day of troubleshooting we had found the

source of the rover's odd behavior: tape

on the

as the rover

team spent

the lander system test ended, the rover

running the rover around in the chamber before

pattern

it

As long

light.

in

fine.

to avoid a nasty sunburn), the

The ghost hazards were gone.

any way. But the rover cameras had

sensitive only to the laser stripers'

narrow

Unfortunately, the tape happened to be extremely

— almost black—when viewed in the infrared, so when the rover's

dark

laser stripe

landed on the tape, almost none of the light was reflected back

to the rover's camera.

When none

of the expected laser light was visible

to the rover, the hazard detection software

the event: "Drop-off Detected."

And

had only one way to interpret

so the rover had run

away from

nothing.

Would there be anything on Mars The planetary

geologists

that could cause a similar problem?

working on Pathfinder were confident

would find nothing as dark as the tape our landing for

it.

site

we had

that

we

inadvertently used. But

proved darker than anyone expected, we'd

now be

if

ready

TWELVE

TWO SPACECRAFT

Despite the bias Pathfinder

was liably,

to

was not

to prove

show

of the rover team, the primary purpose of

we

to place a rover

on Mars.

Instead, the mission

could get a payload to the surface cheaply and

that the

MESUR

to create a global

re-

program's armada of sixteen landers

And

could be accomplished for the estimated budget.

MESUR was

MESUR

network of science

since the intent of

stations,

it

required a

lander that could handle a variety of terrains.

But MESUR's days were numbered. Even a $1 entire

network of landers on Mars began

agement.

The

to

billion price tag for

seem too big

to

MESUR was cancelled. rechristened Mars Pathfinder

went

on.

The low-cost landing

approach being implemented could potentially be employed for other ture

an

NASA man-

Mars missions, MESUR-like or

ments could be mounted on the

fu-

not. Furthermore, science instru-

lander, so the first lander mission to

Mars

in over

twenty years would not only prove the landing system, but also

provide

new data to the

science community. Matt

pointed Project Scientist. this

He would

In the history of the space

ders destined for the

Moon

that— within the confines of —useful science was done.

see to

"technology demonstration" mission

program

Golombek had been ap-

it

prior to Pathfinder,

all

U.S. lan-

or Mars had relied on "powered descent" to

SOJOURNER

146

the surface, and looked a lot like the Apollo lunar module, with one or

more

rocket nozzles underneath and landing legs

upon which

to stand.

When the Vikings of the mid-1970s reached the vicinity of Mars, they followed the same procedure the Apollo

Moon

years earlier, looping around the planet and firing onboard

thrusters at the right

again

had when reaching the

spacecraft;

fire

moment to

thrusters to begin

sphere (unlike the

Moon)

slow into

orbit. Later,

each lander would

descent to the surface. Mars has an atmo-

its

that Viking could take advantage of: a blunt

heatshield protected the lander while slowing the rate of

atmosphere to a more manageable speed, leased to slow

it

further.

jettisoned. In the final

rockets

would be

cal velocity

Then

the heatshield,

moments

fired to

until a its

slow the descent

might land on top of

to topple over or

damaging

its

parachute could be

task complete,

until, ideally,

made

re-

would be

If

came down

the lander

a rock taller than

payload.

the lander's verti-

contact.

The design worked, but it had its limitations. it

through the

before reaching the surface, the lander

reached zero just as the legs

in a rock field,

fall

You could

its legs,

solve the

causing

it

problem by

building the lander arbitrarily big, but you'd pay for it in mass and volume,

which would rapidly make any mission

prohibitively expensive.

Pathfinder was going to do something different.

The

Pathfinder lander was to be a tetrahedron, or four-sided pyramid.

There were no

legs at

The pyramid would sisting

initially

first

a straight shot

aim

its trip it

be encased

its

in a protective "aeroshell," con-

from Earth. Small trajectory correction maneuvers dur-

would ensure

s,

it

would come barreling into the Martian atmosphere that the spacecraft did not miss Mars,

right at the selected landing site. Protected

to Viking

the spacecraft

would

decelerate

by

and would

a heatshield, similar

from nearly seventeen thou-

sand miles per hour to about nine hundred in only two minutes. chute would then deploy while Pathfinder was

When

descent.

phases of the descent. Dispensing with powered descent

altogether, Pathfinder

ing

and no controllable rockets to slow

of a forward heatshield and a rear backshell, which would shield

during the

on

all,

still

the parachute brought the lander's speed

A para-

falling supersonically.

down below about

250

miles per hour, the heatshield would drop away, and the lander pyramid

would lower itself down

a sixty-foot-long cable

from the backshell.

When

Two Spacecraft

14?

onboard radar detected that impact with the surface was imminent, huge airbags ets

would inflate from each face of the pyramid, and small

on the backshell would ignite, bringing the lander to

ing about forty feet in the

The

cocooned

lander,

air.

in airbags,

would

revealing the tetrahedral lander. There

was

to the ground.

fall

the airbags

was no

cut.

And bounce.

would deflate, again which

telling

face of the

on the ground, but it didn't matter: The lander

flat

As the

self-righting.

a standstill, hang-

At just that moment, the cable would be

Many times. When it finally came to rest, pyramid would end up

solid rock-

sides of the

pyramid would open,

like the petals

of a flower spreading in the morning sunlight, their contact with the

ground would

force the lander into an upright position.

open, the lander's solar

fully

would be At

revealed,

least that

With the

petals

cameras, and science instruments

cells,

and the landed mission could begin.

was the

plan.

Some

people thought the scheme sounded

But the Pathfinder spacecraft design team had accepted the respon-

crazy.

sibility

of making

different to land

prior landing,

it

And the team knew that it must do something

work.

on Mars with

a

budget about one-fifteenth that of the

more than twenty years

before.

W" The design of the Pathfinder lander both helped and hindered the rover.

The

rover

petals.

would

Once

only inches

travel to

Mars while

tied

down

one of the lander

would be

the petals opened after landing, the rover

—the

petal's thickness

airbags presented a challenge.

—above

sitting

the surface. But the lander

They might even be

hazard the rover would encounter during

its

the

mission.

was deployed, the lander would have already

rover

to

most dangerous

By

the time the

deflated and retracted

the airbags, leaving loose folds of cloth around the edges of the petals.

If

the rover drove over airbag material, the airbags might catch in the wheel cleats

and wrap themselves around the

before

it

could even begin

you going said,

its

mission. "Everybody said,

to get off the lander?' "

'Ramps.'" Something

like

Howard

red carpets.

to do. When we sold how we were going to get off."

were going

rover's wheels, ensnaring the rover

the rover

'How the

hell are

we "And we had no idea what we to the lander, we had no idea Eisen remembered. "And

SOJOURNER

148

Over

at least the first year, the details

of the ramp idea remained in

limbo. In various rover documents, the ramps were referred to as "draw-

Somehow, the ramps would

bridges," or "red carpets."

rover, providing safe passage over the airbags

and onto the Martian

Whatever the rover team came up with would have space, survive the

weigh very

same

hostile

unroll before the

to

stow

soil.

in a small

environments the rover would

and

face,

little.

Then one day Eisen heard part of a presentation from an outside com"Some guys from Astro Aerospace were at JPL making a sales pitch for some other program. I pulled them aside, and said, 'Hey I've got this thing, I want to maybe drive this rover off this ramp or something. What

pany:

do you think you can that "curls

feeding

comes

it

it

real nice like a tape

out or letting

a stiff

antenna reel,

up

do?' " Astro

it

had products based on

it

stem material

When you let it go,

measure.

spool out,

a

member." Astro had used

its

by

either

goes ahead and curls out, and

it

be-

stem material before to form

booms on spacecraft. When the stem material was rolled up on a

looked

like a flat

ribbon of metal tape.

When reeled out,

the sides

of the tape curled up, giving the tape a circular cross-section; the stem ma-

now looked like

made it much stiffer than any tape measure. Which was just what we needed if we terial

were going to

The

rover

trust

a long tube instead of a ribbon. This shape

our precious rover to

team chose

mitted proposals.

its stability.

to contract out the ramps.

The job looked risky,

the time

was

Few companies short,

sub-

and there was

almost no mass to play with (only about 4.5 pounds). Astro did send in a proposal, tion.

which included

a videotape

showing

a prototype

in ac-

An engineer held the stowed ramp in place, then let go; the ramp un-

rolled smoothly.

The prototype had two of

tracks for the rover's wheels

Astro's stems with attached

and cross supports

in

alignment of the stems. Astro got the contract.

buying into

between

to ensure the

"We thought we were

a very simple, very straightforward system," Eisen said.

"The design proved to be anything but simple." tems had been designed to deploy from ity.

ramp

The ramps would have

be a yard long to clear the

to unroll in airbags.

Mars

And

weight of the rover over that distance.

Astro's previous sys-

free-flying spacecraft in zero gravgravity.

They would need

to

they would have to support the

Two Spacecraft

To make

ramps

the

stronger, Astro

149

needed to use two stems per

side,

not just one. The only option was to place one stem inside the other. This forced additional careful design to ensure that the

two stems

slid

past each

other, rather than binding.

The

original stainless steel of the

rover's cleated

tracks

was too

slippery for the

metal wheels, so the rover had a tendency to

down

trollably

ramp

slide

uncon-

the ramp. Astro added textured surfaces to the tracks to

give the rover traction.

The

it

didn't unroll,

much energy

just

it

roll to

a suddenly looser coil. Eisen:

When

expand outward.

same time."

directions at the

somewhere along

roll,

became

the way.

If

The

the

the system

It

ramp

solution

all

was It it

was

"We had

so

released, the

expanded

'exploded/ did unroll,

in

all

often buckled

to use Velcro,

tape between the coils of the ramp:

peeling open, without loosening

like

When you loosened your grip,

your hand:

in this contained area.

energy caused the

like sticky

The ramp was

design did not always unroll as planned.

final

a tightly coiled roll of paper in

which acted

The ramp could now un-

at once.

But the Velcro made the performance of the ramp temperature-

And we wouldn't necessarily know in advance what time of the Martian day we would be releasing the ramps. "We had to deploy anysensitive.

where from very

early in the

morning to very

late at night.

That range of

temperatures was 80 °C [145 °F]," Eisen said. So they tested the the full temperature range. It

could either be so

the

ramp

stiff

acted up as

if it

Nylon Velcro didn t work

that the

were missing Velcro

qualified stainless steel Velcro uses,

test

it

Velcro and the other side stainless Velcro.

team ran

These

ramp wasn't

a rover

altogether.

A special flight-

tried

tests revealed

in firm contact

and flipping

it

over.

installed

making one

one

final

this

it.

side

few

The

nylon

worked great.

problem:

with the ground

By

after only a

ramp

model down the ramp

the lander petal onto the ramp, the right off

or so loose that

delivered a prototype of the final flight

rover mechanical figurations.

It

temperatures:

all,

much once you

were running out of options. They

When Astro

at all

at

would work, but wore out

which meant you couldn't

testers

ramp didn t work

ramp over

when

ramp could

If

to JPL, the

in several con-

the far end of the

the rover drove from

twist, rolling the rover

time Astro had had enough: The job

SOJOURNER

150

had proven much more challenging than they had expected. Astro that they flight

had delivered on

ramps and be done with

answer to the

weak

at the

it.

flip-over threat,

Modify the ramp to have

bend

their promises.

They wanted

But the JPL team thought they had the

one that Astro could implement

a weak point near the

point, tilting the rest of the

top.

would when extended out too

the ground,

At

first

it

would be

stabilized

company balked

the

them, "You know,

it

far.

Once

way

until the

a metal tape

the end of the

and the rover could

at the

easily:

Then the ramp would

ramp down

reached the ground. The ramp would "break" the sure

felt

to deliver the

end

mea-

ramp was on down.

safely drive

proposed modification. Eisen told

would be pretty embarrassing if we got

all

the

way to

we started driving off the end and the rover you want we can put a little label on the bottom of the

Mars, everything worked, flipped over. If

rover that says, This view brought to

you by our good

friends at Astro/

"

Eisen described what happened next: "The very next day they had their

down at JPL. They took the prototype ramp that we were working with and made the modifications on the prototype to do the sort of thing that we were talking about. And we proceeded to drive our

best engineer

model over the ramp, and

it

worked very well.

"A few days later the flight ramps were modified."

"Minimize the impact of the rover

the top

on the

lander." This decree

was one of

two or three requirements imposed on MFEX. There were many

reasons to simplify the interface between the two spacecraft.

more demands

The more team

the rover depended

on the

would have

on the Pathfinder team, and the more coordination

to place

lander, the

between the two teams would be be

tested.

And that would

necessary. Every interface

would have

to

take time.

All electrical interfaces

through

the rover

between the lander and rover would pass

a "separation connector."

The lander and

rover sides of the con-

when the rover stood up on the lander's nector would petal after Mars landing, or when a pyrotechnically activated cable cutter sliced through all of the wires when commanded by the lander. either separate

Two Spacecraft

The motivation

151

was primarily

for the electrical connections

to keep

the rover healthy while onboard the lander, and to enable the lander to

command several events necessary to Since the rover

deploy the rover after landing.

would be dormant

the entire seven-month trip to Mars,

—

it

totally shut

would be up

down

—

for virtually

to the lander to keep

the rover from freezing or frying. Early predictions by the Pathfinder proj-

would be

ect indicated that the trip

a cold one.

put a lander-powered heater inside the rover's

two wires.

If

the lander

was operating the

indicator inside the rover to

A lander-powered

wires.

more wires,

as

tell it

cations antenna

With

move

seemed necessary

it

needed

a temperature

the rover

would

devices to deploy the rover's

require yet

communi-

and APXS sensor mechanisms once on Mars. and raised questions of

reliabil-

Layman's urging, the rover team began finding ways to

Bill

to

WEB. That required at least

heater,

wake up

A big connector added precious mass, ity.

it

when to turn the heater on or off. More

switch to

would pyrotechnic

So

re-

wires from the separation connector as the rover's overall design

matured.

The new lander thermal models showed Mars, the lander would be

warmer than

that,

at first

during the

flight to

The heater and

thought.

temperature sensor could be eliminated from the design. The pyro release devices and their wires could be

the rover, so

The they

the lander's petal instead of

more wires were dropped.

rover mechanical

ble for the lander to If

mounted on

mounted

team suggested

the relay

on the underside of the

mounted just underneath

the relay

powered on the electromagnet flip

over and close the

power up the acts

would be

rover's

on the lander

for a

circuit,

it

possi-

petal.

few seconds, the relay allowing the rover's

would

on

until

its

lander

on the rover

one of

when

job was done.

its first

switch." This

the lander

power stopped flowing through

stay

could

own batteries to

up,

switch was in parallel with the reed-relay switch. Now,

relay switch, the rover

When the

on the "computer-controlled power

shut off the electromagnet and

it

affixed to the lander,

CPU. As the rover brain booted

to flip

WEB,

rover's

be activated by a small electromagnet permanently

would

make

a "reed-relay" to

power on the rover without any physical connection.

the reed-

SOJOURNER

152

In January of 1994, the separation connector disappeared completely.

The

best interface

was no

interface at

The challenges of making the lander

from the

on the

The

work drew NASA's

rover. Pathfinder's official mission objectives

rover,

The

all.

which was,

rover

rover had

own

little

away

emphasis

in reality, just a payload.

team saw things its

put

attention

this

way:

We

were building a

spacecraft.

subsystems corresponding to each subsystem on

the Pathfinder lander: power, attitude control, telemetry handling, ther-

mal

control, telecommunications,

propulsion).

viewed It

The

and even propulsion

fact that neither the project

NASA

very slow

headquarters

MFEX this way actually worked to our advantage.

was not

until April 22, 1996, barely four

shipped to the Kennedy Space Center, that

was because fly.

nor

(albeit

it

was becoming

NASA

clear that Pathfinder

The impossibly cheap mission

was because

months before Sojourner

to

took note. Perhaps

was

Mars might be

actually going to

a success. Perhaps

And

A

unlike

it

NASA management was beginning to realize that the public

might be paying attention when the experimental rover payload took first drive.

it

failure

most

its

of Sojourner would be an embarrassing spectacle.

high-profile missions,

toring the rover's development to

no one

at

know whether

NASA had been the

damned

moni-

thing was

going to work!

So on that day

in April there

was another

review. This time the board

members were from NASA centers around the country. We needed to show convincingly that we had designed and built a mission- worthy piece of hardware. In the end the board members were surprisingly satisfied with the state of the rover. They still wanted to know why so much of the rover was "single string." Why hadn't we built more redundancy into the hardware? Were they judging Sojourner by the standard of the far more expensive missions of the past? My thought was that if we'd had more mass, volume, and money three years before, we would have built a more reliable rover.

were given.

We

had

built the best rover possible

with the resources

we

Every space mission has an

Two Spacecraft

153

emblem. These

are embroidered or printed de-

signs that symbolize the mission

pants. Pathfinder

had

own. But

its

interacted with Pathfinder, but ting to the rover

team

and

that

we

we

its

goals,

and

MFEX was

tie

together

an independent

weren't Pathfinder;

have our

its

it

partici-

entity.

We

seemed only

fit-

own patch that marked our sepa-

rate identity

Early on,

Howard Eisen had designed a patch for his mechanical,

He proposed

mal, and mobility subsystem.

a variation of this,

ther-

with a

three-quarter view of the rover against a triangular background, as the of-

Sojourner patch. The design was simple, clearly represented the

ficial

and was noncontroversial.

rover,

Some

It

was accepted.

of the rover team, particularly the systems guys and the control

and navigation subsystem, wanted

we were working on something project.

more personal

design.

We

truly

And

the only people

worked on the

For months, on and

all

knew

This was a once-in-a-lifetime

We wanted to come up with the best mission patch

ever seen.

had

a

special.

anyone had

who would get one would be

those

who

rover. off,

the patch

was

a topic of conversation during

the lunch gatherings of rover engineers in the JPL cafeteria. In contrast to the official patch,

we wanted something that made the rover appear more

marauding monster

like a

truck, rolling over anything in

its

path, rather

we were actually constructing. Art who was a professional animator. A few months later Thompson showed up at the lunch table with his

than the microwave oven-sized vehicle

Thompson

said

he knew a guy

friend's first sketch.

mean all

that

its

you could

bugged-out

The

front right

see of the

eyes.

The

rover looked mean.

wheel had

clearly just

poor creature were

rover's

It

its

looked huge.

arms and

a

its

lasers, in reality invisible

and carefully aligned, were instead shooting every which way rays.

was so

crushed a Martian: About

still-struggling

hazard detection

It

like

death

Everyone got a kick out of the drawing. Someone suggested adding

Martian mother pushing a baby carriage, madly fleeing the

invader from Earth.

six- wheeled

SOJOURNER

154

I

could already imagine our first day on Mars.

Control.

The

over and

zoom

would be there

press in

too.

A

We would be in Mission

news cameraman would pan

on the rover team patch embroidered on our jackets.

The Director of JPL and the NASA Administrator would see it was transmitted to millions of people around the world. I

suggested

we

leave out the Martians.

Months passed

again.

Pathfinder Microrover bigger.

The

first set

Pathfinder launch,

The patch was

the image as

The design evolved

further.

Team" now surrounded

The words "Mars The patch got

the rover.

of patches was ready only a few months before the

more than

the largest,

a year after

we had started talking about it.

most complex embroidered design any of us

had ever seen. The embroidery company didn't even want the job of making them: Each patch hours.

would tie up an automated sewing machine for four

Thompson and

made. They were

I

pricey.

collected

At

first

just too expensive." Later they in the halls,

money

to get a full set of the patches

a lot of people chose not to

wearing the patch on a jacket:

one? Gotta have

it."

They just framed

it

"Is it

too

to their kids someday.

it

"It's

on the

wall, or

put

it

Can I still get patch on a jacket.

late?

Some team members never put the and put

buy one:

would see one or two rover team members

in a safe place to give

THIRTEEN

TRIAL BY CENTRIFUGE

The

SIM's

been destroyed."

"What?"

was dumbfounded.

I

morning, and the

first

I

had

just

thing Brian Wilcox told

come

into

work

that

me left me wondering

—Marie Curie—had been undergoing cen-

what to do next. The SIM rover trifuge testing,

and

of yesterday everything had been

as

fine.

What had

gone wrong? Wilcox told me more: He wasn't really sure that Marie Curie had been completely destroyed, but he had the impression

it

been irreparably damaged, or would be out of commission

ther

many months was

that

that

that

it

might

that anything told

what was

him to keep really

accelerations

if

that

at

What he

around

did

ei-

for so

know for sure

sixty gravities,

and he

rea-

had happened, there wasn't much hope

had survived. his speculations to himself,

and went off to

find out

going on.

Centrifuge testing was a

When

have been.

something had come loose

soned, quite logically, that

I

as well

had

we

the Delta

means of subjecting the rover to

expected II

rocket

it

the

to experience during launch

lifted off its

Pathfinder spacecraft and the rover

it

pad, and again at

carried

would

feel

same kinds of and landing.

Mars

arrival, the

many

times nor-

SOJOURNER

156

mal gravity The

down by

had been going on

testing

Most of Wyle's

the airport.

doors, often covered

by an awning

at

Wyle Labs

test facilities

in El

Segundo,

were located

out-of-

equipment from

that barely protected

the elements. But in Southern California, those elements were not so severe,

and you could get away with it. To reach the

down

and about twelve

thirty feet in diameter walls.

The

several asphalt-covered alleyways.

The concrete was marred

other customers'

tests,

had broken

was over

centrifuge pit

with thick concrete

feet deep,

where

in places

you had to walk

test site

items, presumably

free of the centrifuge,

from

been flung out-

ward, and slammed into the walls. During preparations for the rover

some team members would where the destruction of

The

centrifuge itself

play the

a piece of

game of

test,

identifying the locations

hardware had caused a gouge.

was T-shaped and about five and

a half feet high.

The payload to be tested would be mounted on the end of one fifteen-foot arm of the T, while a weight would be slid out and clamped down on the other

arm to provide

was spun

As the centrifuge

a counterbalance for the test item.

up, the object

on the end of the arm would be subjected

higher and higher acceleration, mimicking flight conditions.

wanted

(or

would be allowed)

spinning: If

the rotating

No

to

one

to be in the pit once the centrifuge started

you pressed yourself up

arm might just miss

against the concrete wall, the

end of

arm would be

you. At top speed, the

swinging past more than twice per second.

The source of power diesel schoolbus tal

for the centrifuge

engine in an alcove on one side of the

drive shaft connected the diesel engine

The

drive shaft

was painted

any observer when

it

This did not look

centrifuge

pit.

like a

began to

loud and diesel fumes

would never be

was what looked

filled

the

it

The horizon-

would be obvious

When it was running,

the engine

to

was

pit.

like a place for flight

directly

an old

and the base of the centrifuge.

barber pole so

spin.

pit.

like

hardware. The rover

itself

exposed to the dirty outdoor environment of the

There was

a

cleanroom

Marie Curie had been locked

down

facility

to an

next door.

aluminum

It

was there

that

plate in exactly the

same manner that it would be stowed on the Pathfinder lander petal. Oneeighth-inch steel cables held the rover to the plate in three separate locations. Restraining

hooks held each wheel

in place, so that the rockers

and

by Centrifuge

Trial

bogies could not against

Once

all

it

itself

The

fastened to the plate.

assembly was then moved out of the cleanroom and carried down-

entire stairs

sensor head tightly

of these tie-downs were installed, a clear Plexi-

box was placed over the rover and

glas

in

A final restraint held the APXS

shift.

cradle.

its

15?

and past the

diesel

could then be

engine to the centrifuge. The box with the rover

mounted

in

any of several orientations on the end of

the arm.

This was to be a six-axis

Since the forces the rover

test.

withstand during the mission could

must be

tested in

all six

rover were sitting

on

front-end up (sitting

come from any

on

its tail), left-side

Curie to "qualification"

down (as if the

wheels up (upside down), front-end down,

As with the other environmental test levels,

acceptance" levels only.

"flight

direction, the rover

representative orientations: wheels

a table),

would need to

tests,

down.

right-side

the plan

was

to subject

Marie

while Sojourner would be exposed to

"QuaT

actual flight conditions; the idea

down, and

was

were much more severe than

levels

was

to prove the design of the rover

sound, and would withstand the actual conditions with plenty of margin. It

was

also

the rover.

understood that qual testing would eat into the

If a qual-level test

were continued

chance that something would break.

When

Curie onto the centrifuge in a week's time, half the total acceleration. This level

the flight unit, without reducing

its

total lifetime

indefinitely, there

it

would

was

a

Sojourner followed Marie

would be subjected verify the

to only

workmanship on

remaining lifetime. The com-

effective

bination of tests on Marie Curie and Sojourner

would

give us the confi-

dence that the unit that actually flew to Mars was ready for the rigors

would

After days of setup, Eisen and his

before calling

with

its

it

They had time

The

nose pointed

at the

be tested

7,

1995.

ground, and -x, with It

w as T

It

was just over two

thir-

orientations

was experiencing extreme gravits

held together the

The next day they went through

final axis to

started centrifuge runs with

to test the rover in

a day: +x, as if the rover

the sky. Marie Curie did just fine: to.

team

on Tuesday, November

teen months to launch.

signed

it

face.

the Marie Curie rover

ity

of

good

nose pointed up

way

it

had been

at

de-

the remaining four test cases.

with Marie Curie

effectively

hanging

upside down. This meant that the rover was in a solar array-out configu-

SOJOURNER

158

ration,

with the wheels toward the center of the centrifuge. During

test run, as

with the others, the centrifuge would be run up to the

on

cation level of 66g's, or sixty-six times the force of gravity

would need

achieve this level of acceleration, the centrifuge

ing at nearly 130 revolutions per minute. This run

and

day,

in fact

tests.

When

team

told

was intended

To

Earth.

to be revolv-

to be the last of the

Allen Sirota asked for status in the late afternoon, the test

him by phone

that they

was complete. But during the team

qualifi-

to be the last of Marie Curie's centrifuge

would be done soon, and he could Marie Curie centrifuge

test

run, at around 52g' s, something let go.

The

safely state in his daily status report that the

test

was

this

didn't

know

right

away

exactly

what had happened, but they

were sure something had moved that wasn't supposed to. assembly was removed from the centrifuge after the

what was wrong. The

front left

When the rover

test, it

was obvious

wheel had somehow come

free of

its

"cowcatcher" restraint and slammed into the underside of the solar array!

The

steering actuator

was

actually

Several of the solar cells

embedded into

the panel.

on the array had been cracked by the

colli-

And when they checked the state of the batteries, they discovered that for some unknown reason, Marie Curie was powered on. How had that happened? Had the reed-relay switch, designed sion

between

parts of the rover.

to let the lander turn

immediately

power

on the rover during the long cruise through space and

after landing,

to flow just long

somehow "bounced"

enough

for the rover to

during the

boot

itself

test,

up?

But the rover team had a more immediate concern: Just

had Marie Curie been hurt?

allowing

how

badly

A cursory external inspection would not

an-

swer that question. Electronics, power, and telecommunications subsys-

tems

all

had to be checked

the running gear

was

out.

The

structural integrity of the

also in question.

The team

by the end of the next

at

Wyle packed up

The

rover and returned

it

uation of the state

of the rover would begin Friday morning.

In the

to JPL

WEB and

day.

the

rigorous eval-

meantime, Sirota and others struggled to develop

a recovery

plan in response to this blow to the always success-oriented schedule.

They scrubbed the Sojourner centrifuge test. They would have to fall back to using the

month

SDM for the next set of system tests with the lander, barely a

away. And, assuming that Marie Curie's electronics were

undam-

aged, they

would use those

159

by Centrifuge

Trial

APXS

electronics in

noise testing, even

if

Work would

Marie Curie were partially disassembled for refurbishment. continue on Sojourner.

The engineers

in

the cleanroom were snapping at each other. "Lookyloos"

were showing up to see the wreckage. But the Assembly, Operations (ATLO) team there

who was

Test,

members were having none of

and Launch

that:

Anyone

not doing something obviously important was being told

The ATLO team did not want any extra wounded Marie Curie. They didn't yet know how bad

to leave or justify their presence.

personnel near the the

damage was, and they were very aware

that until they did, the future

of the entire rover mission was hanging in the balance.

When Jake thought

it

was

vious signs of

Matijevic toast."

It

first

damage were

about the inside?

saw the rover

cracks in

What might

some of

The

from Wyle,

The only

the solar

cells.

money to make Marie

"I

ob-

But what

have happened to the wiring, the

board, the sensors, and the integrity of the ing to get the

after its return

wasn't that the rover looked so bad.

CPU

WEB itself? Where was he go-

Curie functional again?

inspection of Marie Curie began.

Some

of the engineers argued that

power up the

rover.

They were

under high acceleration,

its

If that

cleats

were

to even

when the wheel had come free had rammed into the cabling that had

afraid that

been routed along the underside of the aging the cables.

would be too dangerous

it

so,

and

if

solar panel, puncturing

and dam-

unintentional shorts had been cre-

ated between power- and data-carrying wires, then turning on the rover

could cause currents to flow where they were not designed

burning out a half

many of the components on the

hour of visual inspection, no

to, potentially

electronics boards. But after

signs of cable

damage had been

seen.

Continuity tests were performed on individual pins in the connectors on the ends of the cable to see

if

there were any shorts

between

wires.

None

were found. There were only those cracked

cells

on the

During the assessment that followed, the and the inside of the

WEB was examined.

solar array. solar panel

was removed

Rover software engineers Tarn

SOJOURNER

160

Nguyen and Jack Morrison ran the search for further damage turned up

low-level software diagnostics.

The

nothing.

We had been lucky. By the end of the day on the following Monday, had dissipated.

wrote

we embarked on

"Today rapidly

Sirota

.

.

.

The SIM

the sense of disaster

in his daily email report to the rover

the road to recovery, and

we

team that

are healing quite

vehicle should be functional again

by Wednesday."

A dozen solar cells on the panel had been cracked in the accident. Surchange in power output of the

prisingly, the

unnoticeable, with a

maximum

strings of cells

was almost

degradation of 5 percent in one of the

strings.

Within a week of the Marie Curie centrifuge

failure,

the plan to do So-

journer centrifuge testing had been reinstated. The team was confident

same failure could not occur at the 33g level of the Sojourner test. While it was easy to miss the fundamental point in the scramble to rethat the

damage

cover from the

intended purpose: test,

It

to Marie Curie, the centrifuge test

had served

its

was always preferable to encounter a problem during

when no

rather than in flight,

correction

was

feasible.

As we relaxed and proceeded to repair Marie Curie and get back on track, the

rumors

circulated throughout JPL that the rover

stroyed in testing. People I

was going

do now,

to

face

on

a

were

as if the mission

was

okay, but they

bad

situation.

that everything

good

ran into would ask about

I

I

seemed

it,

over.

to think

I

I

had been

de-

wondering what

would

tell

was trying

them

to put a

had too much to do to worry about the

rumors.

While we implemented the Marie Curie recovery plan, one discovery

made

after the centrifuge test failure,

cerns,

began to increase

the centrifuge test fully

shut down.

The

in

but

initially

low on the

list

of con-

prominence: Marie Curie had gone through

powered up! The rover just did not want

control and navigation guys could not

to stay

come up with an

explanation for what was going on. Nothing in the electronic design could

be found that would cause

a

software seemed likely either.

could unexpectedly turn

month

trip to

spontaneous wakeup.

The problem was

itself on,

then

it

And nothing

in the

a serious one: If the rover

might do so during

its

seven-

Mars, arriving with dead batteries. Without batteries,

we

Trial

could never operate the traverse

by the rover into

APXS a

161

by Centrifuge

instrument

at night,

shadow would shut

it

and any inadvertent

down.

We were now rushing toward a December 15 on- time delivery of the Sojourner rover to the lander. We wanted Sojourner to be as flight-ready as

it

could be, although

we knew

that

some unfinished rework items

would remain. But once we handed Sojourner over

would be mounted on

We

during lander environmental testing.

hands on Sojourner again for journer as

would

the lander, and

to the lander team,

it

travel along as a hitchhiker

would not be

able to get our

many months. So we functionally tested So-

much as possible in the limited time. Morrison and Nguyen cal-

ibrated the hazard detection system, navigated the vehicle to waypoints,

demonstrated rover stand-ups, and

mand at least

tried out every possible rover

com-

once.

The assembly and

test

of the rovers had become a complex dance as

boards were removed, modified, and reinstalled, sensors were calibrated,

and the teams

split their

time between two vehicles in different states of

readiness. Sirota's daily "microschedules" orchestrated the flow of activities

out of which the microrovers matured. In early

December

the

ATLO team took Sojourner back to Wyle for a

toned-down version of the centrifuge a

month

earlier.

tests that

had harmed its

This time, the rover survived

its tests,

sister

only

apparently un-

scathed.

^^ The day after Sojourner had been returned from centrifuge testing, Jack

Morrison ran through a sure that the rover was

the

series

of checkouts in the cleanroom to

none the worse

APXS Deployment Mechanism,

or

for wear.

make

He found a problem with

ADM. The ADM was

a simple ro-

bot arm on the back of the rover, powered by a single motor. Once on Mars, the rover could place the

APXS

sensor against rocks or

soil

by

ex-

arm out and down. Morrison reported to Sirota that the deployment mechanism would only go one way. You could deploy the APXS, but the motor would not operate to retract it. Sirota's reaction to the news was unexpected: He smiled. The description of the symptom told Sirota exactly what had failed. The only way for the deployment tending the

SOJOURNER

162

mechanism to move

in just

one direction was

if

ure in the "H-bridge" circuit controlling the eleven H-bridges

motor. These

on the

circuits

rover's

ADM

had been

a

FET fail-

motor. There were

electronics boards,

one

for each

used FETs (Field-Effect Transistors) to switch the

power for the motors on and caused the

power

there

off,

and control

their direction.

So what had

FET to fail?

At the next meeting of the

ATLO team,

Sirota

mentioned the trouble

ADM. Howard Eisen expressed surprise. "It worked just fine when we ran the motor down at Wyle." Something triggered in Sirota's head. What Eisen had said was somehow important. Wyle Laboratories with the

The mechanical

was the place where they had done the centrifuge

testing.

team had done most of the setup

which required stowing

for those tests,

it down to the aluminum plate that simulated the down the rover required driving each wheel indeget just the right tension on the wheel cages. The APXS De-

the rover and locking

lander petal. Locking

pendently to

ployment Mechanism

also

had

to be operated to put the

simulated flight configuration so that just the built

way

it

would be on the way

Ground Support Equipment

to

it

could be tied

to Mars.

it.

There was

What about

its

to the plate,

power each of the motors

a switch for

in turn,

GSE was a box with

each steering motor and

each drive motor, so you could run them forward or backward toggle switches.

in

The mechanical team had

without having to power up the rover brain itself. The ten toggle switches on

down

ADM

the eleventh motor, the

at will.

ADM?

Ten

Sirota re-

membered that he had seen the GSE to operate each of the motor drives, but he had never come across equipment to drive the ADM. The connectors for the ten mobility

motors were easy to get

to,

located

on the

front

WEB. But the connector for the ADM motor was on the back of WEB, and the ADM itself tended to get in the way. It was suspicious

of the the

that the only

motor giving them problems was the same one

knowledge, had no

GSE box

to

run

it.

How

that, to his

did they operate the

ADM

motor? Eisen told

him

that he

would just connect the output from

a

power

supply directly to the motor leads themselves by poking sharp probes

through the coating material, and the motor ran just

fine.

Trial

The

by Centrifuge

pieces of evidence in Sirota's

Sirota asked the next question, but

the

163

mind came

together.

It

clicked.

he knew the answer. "Did you remove

ADM connector from the rear bulkhead before applying power to it?" mind was working

Eisen's

too,

somehow

perceiving a threat. "No.

Why?" "Well,

motor. right

bias

if

you put power on those

also goes

It

back to the H-b ridge. The

you put on

it

course, eventually,

The

wires,

FET

doesn't just go out to the

isn't built

to take the large reverse

every time you applied power to the connector." So of it

had

failed.

were going on around the

lights

it

through the connector onto the rover power board,

table.

The engineers on

Eisen's

team understood the explanation. And they believed it.

And rover's

power

the

power

bus,

didn't stop at the FET.

and woke up the

rover's

The power

fed right into the

CPU. Even though Eisen was

feeding power into the rover for only the few seconds

took to deploy the

it

ADM onto the plate, that was enough time for the CPU to boot up and do what

it

C on-line. Once that hapwas commanded to shut down

always did, bring battery strings B and

pened, the rover could stay on until again, or

its

batteries

it

went dead.

The following Monday they opened up Sojourner to take a look. The for the ADM motor were on the top side of the top board in the

FETs

WEB. There where

it

covering

other

it

was: a blackened spot on the board surrounding the FET,

had overloaded, overheated, and burned the coating material it.

But once again, the rover team was

damage

lucky, for there

to the board.

Once he understood what he had done, Eisen was the rover

was no

team was just

chagrined.

Most of

relieved that they finally understood the

mecha-

nism by which the phantom wakeups had occurred. Mysteries that did not yield to reason and analysis were disconcerting. Knowing that error could fully account for the observed evidence

human

was reassuring

to the

engineers. It

was

satisfying to the rest of the

mistake. In the prior rors, or for acts that

two

years, Eisen

team

to see Eisen publicly admit a

had lambasted others

for their er-

He was

often called

Eisen had decreed were errors.

SOJOURNER

164

the "Teflon engineer" because blame never stuck to him: ful at deflecting

it.

Immediately ness

He was very smart,

and he was driven to be

cal engineer,

made him

after the

easier to

riod of his humility

an exceptionally capable mechani-

right

about everything.

mysterious wakeup incident, Eisens contrite-

work with. Had he learned

was

He was too skill-

short-lived, only a

a lesson?

few weeks. Eisen was

warrior momentarily stunned by a blow Soon he shook fully recovered,

ready to do battle again. Eisen was a key

rover team:

We

job done, so

we put up with the

needed both

No. The pe-

it

off

like a

and was

member

of the

and expertise to get the

his aggressiveness

day-to-day frictions.

As we pushed toward the goal of handing the Sojourner rover over to the lander, the delivery date receded

ready to receive the rover into for the

handover began to

slip

its

from

in the early

Sojourner was

moved

day by

keep their

day.

1,

team

The

date

used the addi-

an ongoing noise problem with the 1996, the carefully

boxed

the few hundred feet to the Pathfinder cleanroom

own

where the lander

folks

had been working long

assembly, integration, and test operations

track. Despite their efforts, the lander

schedule.

The

rover

morning of January 23,

in the building next door, shifts to

The lander simply was not

integration and test activities.

tional time to attempt corrections to

APXS. Then,

us.

on

team was twenty days behind

Somehow, they had to get those twenty days back.

On February

the Pathfinder project held a review of the rover documentation and

accepted delivery of Sojourner. With that acceptance,

status,

and

Sirota

announced the end of the rover implementation phase, and the

start

officially

of integrated spacecraft testing and rover

tion. It

was

resolved,

a distinction

and the

already revealed

first

itself.

drawn in the

sand:

flight readiness prepara-

Old rover issues remained un-

threat to integrating Sojourner with the lander

had

FOURTEEN

CAN WE TALK?

walked through the card-key protected doors into the Pathfinder control area.

I out of

Computer

names of

workstations, assigned

Star Trek for easy identification,

test

characters

were everywhere. Engineers

were wearing jeans and headsets with push-to-talk microphones. Some of

them munched on breakfast while studying At the

far

other world.

was

of

air

their screens.

low window looked out

a long,

The high-bay cleanroom of the

huge room, with

a

river

end of the room

ters cleansed the air

on

its

way

A

room from one end to

particles too small to

uniform

the other,

be seen. High-efficiency

fil-

The

at-

to another transit of the high bay.

mospheric pressure inside the high bay was so any breach of the room's seals dirty air to seep in.

Spacecraft Assembly Facility

a ceiling at least forty-five feet high.

flowed constantly across that

sweeping up suspended

into an-

higher than outside,

slightly

would cause clean air to leak out, and no

You entered the high bay only by

airlock.

While

trapped in the airlock, you would be subjected to a high-pressure

shower" that removed any loose

The form

you might be carrying with you.

assembled Pathfinder lander rested on a mobile

few yards from the window through which

a

lander

partially

dirt

I

saw the spacecraft

mounted

in

its

"air

cruise stage

I

plat-

peered. Beyond the

and backshell. Each one had been

own aluminum framework, which

could be rotated and

SOJOURNER

166

tilted for

easy access by the lander assembly and integration team.

high bay had been built to accommodate

much

merging elements of Pathfinder were dwarfed

The engineers on more formally, not in

larger spacecraft.

in that

The The

volume.

the other side of the thick glass were dressed far

and

suits

ties,

but in aptly

ff

named bunny

suits":

lightweight white smocks and pants, white hairnetlike caps over their

heads and white booties over their shoes.

Human

beings are

dirty.

They

are shedding

flakes of skin, moisture. Particles they've

clothes with them. ticles its

per minute.

The bunny

been carrying around on

suits

that remained. Partly, the intent

over 2 million par-

were there to protect Pathfinder from

was

bay carried away most of the dust

to

minimize biological contamina-

we

This hardware was going to Mars, and

nate any potential Martian planetary protection policy

was

their

shedding to only a few tens of thousands. The

carefully designed air flow of the high

the spacecraft

the time: hair scales,

An average person easily generates

creators, cutting the

tion:

all

life

want

on

to contami-

with microbes from Earth. NASA's

would not allow Pathfinder

clean. Just as important,

settling inside a fuel line or

didn't

we

a circuit board.

didn

to launch unless

want any

t

particles

A thruster that failed to fire,

or an electronic component that shorted out, might end the mission before

it

started.

Humans

also carry electric charges, static electricity

stroy electronic components.

The

fabric

of the bunny

which can

suits

de-

contained a

grid of conductive thread to dissipate such charges. Conductive straps ran

from the wearer's foot to the underside of the white

booties, to contact

the conductive floor of the high bay, preventing charge buildup.

Also in the high bay was Sojourner.

had been delivered Operations Mode"

hours before.

just test,

the

Pathfinder lander. This test the interactions of the

two

first

It

was January 23,

Now

it

was time

1996.

The

for the "Surface

joint test of the flight rover

would be the

first

rover

and the

chance to practice some of

spacecraft planned for the

first

day on Mars.

The flight lander would send command sequences to Sojourner, which would respond with telemetry data. For weeks prior, I had worked with lander engineers to define the detailed procedure we would follow today. Every

activity involving the Pathfinder flight

hardware required an ap-

We Talk?

Can

proved, signed-off procedure. This

cedures were our

of

activities

first

was the only lander we had: The

defense against

between now

16?

human

pro-

error during the myriad

and having a functioning spacecraft on

way

its

to Mars.

Thompson and

Art

other rover team members, wearing their

own

bunny suits, watched over Sojourner in the high bay. I remained in the control area, ready to send

test

commands through the lander to the rover. We

were separated by less than twenty feet. Yet the only way to communicate with

Thompson was

visitors

and inmates

talking

by telephone.

The

test

either via the

communal headset

in prison, staring at

voicenet, or like

each other through the glass and

conductor slowly stepped us through the procedure. Con-

firm the test-cable connections to the lander. Check the voltage levels on the

power supplies. Apply power to the lander bus.

Finally,

was

the lander

up and running. The rover telecommunications guys had previously and

livered

de-

lander-mounted rover radio and antenna. One

installed the

of the engineers at a computer workstation sent the lander

command

"MODEM_POWER_B"

radio; the

next

to turn

on the lander-mounted rover

command was "ROVER_WAKEUP." The

livery box,

was powered

rover-lander radio link.

We

up.

waited. But

rover, fresh

out of

no data flowed

With no commands from the

its

de-

across the

lander, the rover

made no moves. The lander and rover generated only error messages. The test had failed. Or rather, it had never even begun. Lander-rover communications wasn't functioning.

The team dropped test

into

debugging mode. Almost immediately, the

conductor wanted to skip the rover

test

and move on to the next pro-

cedure.

The

rover-lander link should have worked. Art

rover side and Glenn Reeves

weeks

in the testbed, testing

The

link should

on the lander

side

Thompson on

the

had already consumed

and debugging communications.

be working!

The lander and rover teams had each spent so

much

time

in the past prov-

ing out their pieces of the radio interface that each had the

same

reaction:

SOJOURNER

168

"It

must be your problem. What's wrong with your

side of the interface

again?"

"By

this point,

we were

wanted the other guy

to

all

show us

from Missouri," Thompson that his interface

"We

said.

was working

correctly

before we'd consider that there might be a problem on our side." At the

end of the two days allocated to the surface operations test, the lander and rover were

still

not speaking. The lander integration schedule was

The engineers working

the communications problem

treat to the testbed to continue

tight.

would have

to re-

debugging, while spacecraft integration

moved on to the next critical activity. The communications failure looked suspiciously like a problem Thompson and Reeves had isolated in the testbed months before. At the time, they

had traced the problem to a design flaw

electronics board.

been

fixed! All

The bug had been

in a chip

on the lander

a big deal in September.

But

it

had

of the computer boards had been sent back to the manu-

facturer

and the bad chips had been replaced with new ones that

have the

flaw.

didn't

Why would the old symptoms suddenly appear again now?

Within three days of the

failure,

the lander

team had checked

records and confirmed that the computer board installed

their

on the lander

had never been shipped back to the contractor to get the bad chip placed. Every other board, including those in the testbed flight

computer board intended

stalled,

already had the

to

that

one

that there was, for now,

It

would take

disassemble the lander enough to get to the board.

a full

in-

week to

The lander team was

weeks behind schedule. The soonest they would have

replace the

final

now

we understood the problem. The bad news no way to correct it. Opening up the lander

swap out the bad board was not an option:

already

and the

fix.

The good news was was

to eventually replace the

re-

a

chance to

bad board would be mid-May almost four months away. And

even that date was in question, since one of the easiest ways to make up schedule would be to delay installing the just a

final flight

few weeks before shipping the lander

Center.

computer board until

off to the

Kennedy Space

Can

This

was

And

as

not the

first difficulty

many

with

We

with the rover communications subsystem.

other elements of the rover's design,

and her team had broken the rules to get

Most

spacecraft

Donna

Shirley

this far.

communications systems were designed to send

nals over long distances. If a spacecraft

those distances

169

Talk?

would be measured

in

were

in orbit

sig-

around the Earth,

hundreds or thousands of miles.

If

the spacecraft were traveling to another planet or through deep space, the distances

would be

would never need help.

The

space;

it

millions or even billions of miles. But Sojourner

to

communicate with the Earth,

rover's radio wouldn't have to

had

just

not without

at least

be heard across interplanetary

to talk to the Pathfinder lander, at

most only

a

few hun-

dred yards distant. The lander would be Sojourner's communications lay

The lander would have more of everything to do

from

its

more power

the job:

larger solar arrays, a "High-Gain" antenna that could be pointed

at the Earth,

own

re-

and

huge computer memory

a comparatively

to store

its

data and that of the rover.

From

the start of the

rover's radios should

This was not

be

a

MFEX

effort, Shirley

had decreed

that the

purchased commercial product.

how JPL

did things. For each spacecraft, the

JPL com-

munications section would study the mission requirements, determine the necessary specifications, then either design and build the tions system in-house, or contract

enced with

flight

it

to an appropriate

communica-

company

experi-

hardware. Either approach would likely cost several

million dollars. Shirley

knew her

$25 million total rover budget was stretched

tight.

She didn't think she could afford a huge chunk of that budget for designing and building a custom flight-qualified communications system. Lots

modems that allowed two And from the standpoint of

of companies produced radios, and even radio

computers to pass

digital data

between them.

talking to each other, the lander and the rover

Radio

modems on

must be an is,

existing

were just two computers.

Earth could converse over miles of ground. There

commercial radio that could meet the

rover's needs as

or slightly modified. Shirley preferred spending her limited

the robotic aspects of the rover that

Even

as the Pathfinder Project

made

it

money on

unique.

Manager was attempting

to pressure

SOJOURNER

1?0

Shirley into using a tether instead of any radio at

a

all,

communications

engineer on Shirley's team was doing an industry survey to locate commercial radios that might one day serve as Sojourner's link to tors. its

Two

rover radios

would

fly to

twin installed on the lander. By the

didate

had been

Once

ple of those

late

Motorola

was chosen,

were sent down to Building

opment Model Motorola

identified: the

the Motorola radio

rover took shape

modem was part of

it.

mounted on

Mars, one

summer of

its

opera-

the rover, and

1992, the best can-

RNET 9600 radio modem. several 107.

were ordered, and a cou-

As the

on the gutted

The second

first

Software Devel-

chassis of

radio

Rocky

was wired

4,

one

to the card-

cage that simulated the lander side of the communications system. For the lifetime of the rover development effort, this radio link

preferred

would be the

means of sending commands and telemetry

The JPL communications engineer who had selected the on, leaving JPL for a

radio

new job. Soon thereafter Lin Sukamto came onboard

the rover

team

first flight

project assignment.

as the communications Cognizant Engineer.

It

then proving that they would survive the rigors of the

rover's

antenna and

and temperature. They would its

mate on the

flight

for flight,

environment:

also have to design the

lander.

The radios were only a few hundred dollars that

apiece.

They were so cheap

Sukamto ordered thirty of them. Then Sukamto and her team began

methodical, rigorous screening process to select the "best of the

For

was her

The job of Sukamto's telecommunications

team now became one of repackaging the Motorola modems

vibration, radiation,

moved

a

lot."

two radios to communicate, the receiving radio has to be tuned to the

same frequency as the transmitting radio. Radios depend on tiny components called crystal oscillators to regulate their frequencies.

watches owe their accuracy to their own quartz tals

vibrate millions of times per second.

around the actly the

crystal

remains

As long

crystals,

The

crys-

environment almost ex-

to the next. This allows

radios to consistently transmit and receive at the

two

as the

fairly stable, the crystal oscillates

same number of times from one second

rover radio had

Quartz-movement

crystal oscillators.

same

frequency.

one to control the frequency of

its

Each

transmit-

Can

ter,

We Talk?

one to control the frequency of

maintain their frequencies under

each crystal went up, so did Testing

were

receiver.

its

But the

As the temperature of

conditions.

all

crystals could not

frequency.

showed that if the temperatures of the rover and lander radios communications link worked fine. But if the temperatures

close, the

drifted far

its

171

enough

apart, the frequencies could shift to the point that the

The Motorola modems

radios could lose the ability to talk to each other.

were just not intended

for the

temperature extremes of the Martian envi-

ronment. The custom-built radio systems on JPL spacecraft readily handled wide temperature ranges. Frequency

problem case,

if

drift

would not have been

a

the radios had been designed and built in-house. But in that

of course, the cost

itself

would have been the problem.

come up with a complete solution to the frequency drift issue. As Bill Layman would have said, they had not "killed the problem." They turned to Jim By

late 1995, the rover

communications team had

still

Parkyn, one of the JPL communications technical gurus. solution: install

One "oven."

temperature-compensated crystal

The oven kept

this feature:

the crystal toasty

and so did

The component was

and would not

fit

warm

its

oscillators.

at a fixed

bigger,

inside the existing radio.

drew

The

oscillators

was far too

still

wouldn't

testing.

own

tiny

temperature.

a price for

for the oven,

rover didn't have the

power

packaged with additional

cir-

These temperature-

into the radio, but at least they

available commercially.

late to

incorporate the change into the original rover de-

The question was whether

time. Sojourner

fit

power

extra

cuitry that corrected for changing temperature.

were low power and

its

you paid

frequency. But

to spare, so Parkyn suggested other crystals

compensated

a

to the temperature outside the package, the crystal

inside stayed constant,

It

He proposed

option was a specially packaged crystal, encased in

Whatever happened

sign.

not

the flight rover could be retrofitted in

was already mated with

Pathfinder, undergoing system

We would have to implement the fix, install

it

on Marie

Curie, and

prove to ourselves that we would be able to safely modify Sojourner in the short time the flight rover

would be

in

our hands again before delivery to

Florida.

We bought

oscillators.

Scot Stride, one of the rover telecommunica-

SOJOURNER

1?2

remembered examining

tions engineers, rived.

by-side, a

the oscillators

when

they

"They were huge! They were enormous! Putting two of them it's

almost as big as one of the radio boards.

ar-

side-

And we had to build

board for both of those." The inside of the rover was already

tightly

new

oscilla-

packed. Finding space for another electronics board with the

mounting the board

tors,

to the existing radio

safely in that space,

modem was going to be

the rover already designed and built, there

and wiring from the board

a challenge.

With the

was always the

rest

fear that

of

new

changes might create more problems than they solved. Scot worked with

team

the rest of the "Electrically

to test the effectiveness of the

we got one

side to

work.

It

worked

really well."

frequency stayed locked in over wide variations in "Mechanically,

was

it

new components.

modem

The

radio's

temperature.

a nightmare." Attempts to integrate the oscillators

continued over several months, as the sole opportunity to rework the guts of Sojourner rapidly approached. But a reliable solution eluded ing to implement

on one of those

it

radios

would have been

us. "Try-

really messy.

And we just ran out of time." Sojourner would

fly as

is.

HPf

For

months

after the

bad communications chip had been discovered on the

lander computer board, the rover team used clever workarounds to keep

Sojourner involved in the lander system

testing.

The

eventual resolution

of the problem was anticlimactic. The computer board that would actually fly to

tem

test

Mars was

on the lander in late May. The next sys-

confirmed that Sojourner and Pathfinder communicated

were designed rover

finally installed

to do.

team could

With each

relax a

little

successful

communications

as they

session, the

more.

Jan Tarsala was a JPL communications engineer.

One

day, six

Jim Parkyn

months or so before the Pathfinder launch, he ran

in the

JPL

being engineers in

cafeteria. Tarsala

and Parkyn were

the JPL Telecommunications

into

friends. Besides

section, they

were both

Can

amateur radio operators.

had

first

We

was through

It

when

communications engineer-

left

they went home. They had radios in their blood.

Parkyn was cheerful.

"I've

dustry and I'm leaving the

been looking

mutual avocation that they

their

met. Neither of them completely

ing behind

1?3

Talk?

some

for

some news

Lab orator v."

for you. I've got a job in in-

knew

Tarsala

you know

time. "'Did

was news

also leaving?'' This

got

that Parkyn

had

van Nieuwstadt

that Lin

to Tarsala, but not a surprise. Lin

is

Sukamto

had recently married, her husband was Dutch, and they were relocating Parkyn went on. "You know what

to Europe.

all this

means. You are going

to end up with the Sojourner radio job."

That was

a surprise.

"What

are

you talking about?" Tarsala was

ready up to his elbows with the radios on another space mission.

al-

He

protested that he couldn't possibly take on another task.

Parkyn persisted. "No. Xo. Lin It

would only be

knocking on

is

awav. Let

it

So they

sat

leaving.

is

You

are the heir apparent."'

matter of time before section management came

"They

Tarsala's door.

going to say This to turn

a

are

going to come to you. and they are

your job/ And you are not going to be

me

vou what vou need

tell

down in the

to

and in about

cafeteria,

in

any position

know about

a half

this job."

hour Parkyn gave

Tarsala the technical history of the Sojourner radios. Parkyn Tarsala:

warned

"You're going to land on Mars, and you're going to be

frequency

And you

that situation

are

off-

going to have to come up with a way of managing

and making

it

work." Parkyn described the attempt to add

the temperature-compensated crystal oscillators to the radios. These

would

He had pushed as hard as he could to get the new oscillators implemented, but somehow Lin and the rover team had shipped the hardware without the fix. "Just know have solved the frequency-drift problem in hardware.

what you're getting

into.

You're not going to be able to refuse

it.

Be

fully

prepared for what's going to happen." Tarsala kept

what Parkyn

told

trusted his technical observations. practical side of

Radio with

I

him

to himself.

"I

trusted Parkyn.

trusted his opinion. Parkyn

knew

I

the

a big R.'"

Sure enough, the Telecommunications section manager came to Tarsala a

week

or so after Lin van Nieuwstadt

left for

Europe.

"I've

got a

SOJOURNER

1?4

job for you. You will be working for Sami Asmar,"

who would be the new

Cognizant Engineer for the Sojourner radios. "The whole section

is

com-

mitted to having a successful mission." The section manager turned to Tarsala, pointed at his chest, for

making

this radio

fered to Tarsala, but

and

said,

"But

I

expect you to be responsible

work." As Parkyn had forecast, the job was not

commanded. And thanks

to Parkyn, he

forewarned of the challenges that were yet to come.

of-

had been

FIFTEEN

THE NOISE THAT WOULDN'T DIE

Allen Si rota's father had been an electrical engineer at Ford Aerospace in

New York.

Eventually he

moved the

family out to California, and

got a job at North American Aviation, working on everything from the XI 5 rocket plane to the Apollo

moon

program. Sirota remembered

growing up watching Walter Cronkite on Gemini, and Apollo missions. majors but

—inevitably

it

When

TV

describing the Mercury,

UCLA, he

he got into

—ended up

seemed

tried several

in electrical engineering.

After ten years in the aerospace industry Sirota realized that he wasn't really enjoying the job

to the idea of bility. "I

he found himself doing.

working

at JPL,

thought you needed

He had always been drawn

but had never viewed

a Ph.D. to

sweep

it

as a serious possi-

the floors here!" Sirota later

said in his JPL office.

When

he

finally

number of job

went

in to interview at JPL,

qualifications. "Sure.

I

he was asked about a

can do that" was always

Sirota's re-

sponse.

He was hired. that?"

Sirota reviewed

really

do

immediately

as-

what he had promised. "Can

he asked himself. Well, he would find out.

He was

I

signed to be the technical manager on a small flight project, an experi-

ment

that

would

fly

on the Space

Shuttle.

When

was already behind schedule, over budget, and

in

he

arrived, the project

need of

a "miracle."

He

SOJOURNER

1?6

took these problems later

looked back

time.

He was

at this first

new to

was indeed up

JPL assignment

to the task,

as his favorite project

of

all

the organization to be keyed into the workplace

would he enjoy such blissful ignorance.

But working on rience he

a rover

going to Mars wasn't half-bad.

had gained learning to deal with Principal

was about

smaller flight experiments

With the exception of the lander's

to

be

mine what things were made could, given

it

enough

And the

expe-

Investigators

on

APXS was

the

useful.

IMP camera,

the rover's

key science instrument on the Pathfinder mission because

rocks,

and

and he was given free reign to solve the technical problems. Never

politics

again

too

as a challenge. Sirota

it

would

deter-

When the instrument was placed against

of.

time, produce spectral data

from which the

elemental composition of the target material could be discerned. By

knowing the types and and so on)

in a rock,

relative

abundances of atoms (such

you could learn

The APXS was in two parts: rover, to

side the

be placed against

a lot

the sensor head, which

targets;

a

bits

in-

compiled the raw data streaming

complete picture of the

pended on nine small pieces of curium mounted These

was outside of the

and the electronics box, mounted

WEB, which over several hours

from the sensor head into

as iron, sulfur,

about what kind of rock it was.

target.

The APXS

de-

in the sensor head.

of radioactive material emitted alpha particles that would

strike a target

rock whenever the

the alpha particles

APXS was

bounced back from the

Some

of

Sometimes protons

in

deployed against

target.

it.

the target material were also dislodged, and sometimes the alpha particles excited the target's

atoms to produce X

rays.

Detectors in the sensor head

picked up the pattern of alpha, proton, and X-ray radiation reflected back

from the

target.

in the sample,

By analyzing the

and

their relative

The APXS was to be

pattern,

you could identify the elements

abundance.

a joint effort of the University of

Chicago and the

Max Planck Chemical Institute in Mainz, Germany. Rudi Rieder, from Max Planck, was the APXS Principal Investigator. This made him the lead scientist responsible for

for inclusion

on

developing and delivering the instrument to JPL

the Pathfinder mission. Rieder's counterpart at the Uni-

The Courtesy of

target NASA/USGS

The Blue Rover Courtesy of NASA/JPL/Caltech

The pantograph:

Don

Bidder's

first

high-

mobility vehicle design Courtesy of NASA/JPL/Caltech

Robby: the

first

excursion into the

Arroyo Seco Courtesy of NASA/JPL/Caltech

Tooth: a tabletop rover Courtesy of

NASA JPL

Cakech

fe

Go-For: the fork-wheeled

microrover

\ >..

i:m.

Courtesy of

NASA

Rocky

4:

JPL. Cakech

the

Mars

Science Microrover Courtesy of

NASA JPL Cakech

The photograph from entation to

new

the Viking

employees:

1

lander

I

What might be

showed

Courtesy of NASA/JPL/Caltech

Sojourner on the benchtop Courtesy of NASA/JPL/Caltech

in

my

pres-

over that horizon?

Sojourner in the twenty-five-foot chamber with the Pathfinder der for the final thermal / vacuum test

courtesy of nasa/jpl. caitech

lan-

The

unofficial rover

team patch

Courtesy of Calvin Patton

Sojourner team members in the JPL MarsYard with Marie Curie. Top row, Firenze Pavlics,

left to right:

Hank Moore, Tarn Nguyen, Dutch Sebring, Matt Wallace, Lee Sword, Ron Banes, Howard Eisen, Ken Jewett, Henry Stone, Jim Parkyn,

Fotios Deligiannis,

Ark Thompson, Jack Morrison, Allen

Sirota.

Lin Sukamto (van Nieuwstadt), Beverly Stride

.

Courtesy of

NASA /JPL /Caltech

St.

Bottom row. Brian Cooper, Jake Matijevic,

Ange, Fred Nabor, Andrew Mishkin, Scot

Sojourner checkout

at

Kennedy Space Center

Sojourner joins Pathfinder

at

the

Cape

i

cmatayefNASA

Closing up the lander for the

last

Pathfinder and Sojourner

on

their

way

to

Mars

Courtesy of

NASA

time

courtesy of

nasa

Landing Day: Sojourner and Pathfinder on Mars

Sojourner's

first

lem.

lost

NASA/jpucaiuch

images from Mars: views of the

ward ramp before and was

courtesy of

after

deployment.

Some

for-

data

due to the rover-lander communication prob-

Courtesy tf

NASA JPL

Caltech

The soil

first

rover movie:

six

wheels on

(The yet-unseen rover causes the

ramp

to shift in the early frames.)

Courtesy of NASA/JPL/Caltech

Six

wheels on

Courtesy- of

SASA JPL

Sojourner

soil!

Sojourner touches

down on

the Martian surface.

Caltech

hits

Barnacle

Bill

on the

first

attempt. Camay •tnasa

jpl caUech

image of the lander taken by Sojourner

Sol

3:

the

Sol

5:

the second rover image of the lander

Sol

8:

Sojourner takes a picture of one of

proving

it

first

can be seen in Martian daylight,

courtesy of

its

courtesy of

NASA/jpucaitech

NASA/jpucdtech

hazard-detecting laser stripes,

courtesy of NASA/jPL/caitech

Sojourner bags the rock called "Yogi," as seen from the lander. Courtesy of SASA.JPL/Caltech

A rover-eye

view of Yogi

courtesy of

nasa jpl

coua

During

a soil experiment, the rover

lifting its front left Courtesy of NASA/JPL/Caltech

wheel into the

does a wheelie,

air.

Sol 35: Sojourner ible in the

sits

near the rock "Wedge." Part of the Rock Garden

upper right of the image,

courtesy of

is

sasa jpl catah

Sojourner spies sand dunes behind the Rock Garden.

camtayefHASA jpl abah

vis-

w&m. 'mmmmm-mmmv^

Two generations:

Marie Curie in the cleanroom with the twin Mars Exploration rovers

Courtesy of NASA/JPL/Caltech

The Noise That Wouldn't Die

Chicago was co-Investigator Thanasis "Tom" Economou.

versity of

Economou's

style

was

naturally combative.

mands on us, he was arguing with ing,

up

1??

to a point. Together, they

When he

Rieder. Rieder

wasn't making de-

was much more easygo-

were an ongoing challenge to the rover

team. The interaction between the JPL team and the

was

a classic clash of cultures: flight

APXS

developers

hardware engineers versus university

scientists.

In late

February 1995, Rudi Rieder arrived in Chicago, having traveled from

Germany with the APXS electronics destined for installation inside Marie Curie. From O'Hare Airport he took a cab to the University of Chicago, where Tom Economou waited. Within days they would be putting the instrument through qualification testing

and vibration environments

On

way

the

it

would

at JPL, subjecting

went

in to

slammed pital,

to the thermal

see in space.

to the university, Rieder asked the cabdriver to stop at a

Starbuck's coffeehouse at the corner of South Fifty-Third.

it

While the

taxi

Harper Avenue and East

waited outside with the engine running, Rudi

grab a coffee. At that moment, a fifteen-year-old fleeing police

his car into the rear

but the

APXS

of the cab. The

taxi driver

electronics in the trunk

smiled for the photographers and told the press:

shock test.

It's

been done

went

to the hos-

were undamaged. Rieder

"We don t need

to

do

a

in Chicago."

The Marie Curie vehicle began taking shape in the middle of 1995. So-

journer followed soon behind. By

late

October, Allen Sirota was collecting

APXS electrically wired to the naked Sojourner CPU board on a benchtop. Economou did not like the results: The spectra were test spectra

full

with the

of electrical noise, masking their meaning. They were supposed to

show

as

graphs with several sharp peaks,

range against the

sky.

The

like a silhouette

positions of the peaks

of a mountain

would represent

the

ferent elements in the material being analyzed. Instead, the spectra

muddy; one, as

all

if

dif-

were

of the peaks that should have been there were blurred into

you were looking at the mountains through smoked glass. Noise

SOJOURNER

178

was masking the APXS strument would be Electrical noise

rover electronics. electricity to

signal. If the noise

got bad enough, the

was leaking

into the

APXS from somewhere

APXS

spectra, the noise

action to the

electronics. If this

four

problem was

this

was only

typical,

APXS

slightly

APXS

ing like an antenna, picking the rover. But

where

find out

On November

would

out the

improved. Economou's

Why

can

t

re-

his favorite re-

you?" There were

noise problem

same time

was unique

power supply was not the source of the

cable connecting the

To

filter

instruments included in the payload for the Russian Mars '96

Economou insisted that the the

power should

and destined to become

mission, scheduled for launch at about the

If

were the cause, then ca-

minor modification. But when they

"The Russians gave us good power.

frain:

in the

A likely culprit was the power supply that provided the

operate the

The team implemented

new

took

in-

unable to distinguish the elements in a rock.

useless,

pacitors placed across the wires supplying the noise.

APXS

like to

noted by the

1,

electronics

as Pathfinder.

to the rover.

noise, then perhaps the

and the sensor head might be

act-

up the noise radiated from some other part of

exactly

was the source of the

noise?

more, we'd have to put Sojourner together.

1995, Allen Sirota sent an email report to the team:

announce the birth of the first

power

application of

flight rover

in

"I

Sojourner Truth, as

an integrated configuration.

much more testing lies ahead. To protect Sojourner from any possible harm or contamination, the following guidelines will be adhered to when working on Early test results indicate

no problems

or in the proximity of the vehicle

at this time,

although

." .

.

m By mid-November, APXS testing on the flight rover could resume. Matt Wallace,

from the rover power subsystem team, led the

were

as noisy as before.

testing.

The

spectra

As the rover team continued debugging the problem, Wallace found himself working only with the abrasive

where the instrument

Tom Economou. We wondered

Principal Investigator was. Rudi Rieder

might be

1?9

The Noise That Wouldn't Die

the needed buffer between

Economou and

have some ideas the other scientist didn help,

we were informed

JPL engineers

felt

he was

that

the rover team, and might

When we

t.

in Russia,

like they were under the

gun

asked for Rieder's

and was unavailable. The to get this

problem

fixed;

APXS team to help. Everyone assumed that Rieder was busy preparing additional APXS units for the Russian Mars '96 mission. Whatever the reason, we would not see him

yet they couldn't get the full resources of the

again for months.

Now the push was on to deliver Sojourner to the lander by December 15, 1995,

only a few weeks away. Sirota had created a plan to meet

deadline.

There was

a lot of environmental

this

and functional testing of So-

journer that had nothing to do with the APXS. Continued testing of the

APXS would

threaten the delivery schedule. Sirota decreed, "The

no longer be

noise problem will

the next

month

investigated until after

team turned

the rover

its

attention to

APXS

FUR delivery."

For

making Sojourner

as flight-ready as possible.

The Pathfinder the

APXS noise

Project

Manager had other plans. Tony Spear wanted

corrected before accepting the rover delivery.

With lander

integration running late anyway, Spear offered to extend the rover's deadline to give the

team more time

came otherwise ready instituted

lem

at

an

APXS

to find a solution.

for delivery

"tiger team," led

So when Sojourner be-

on the eleventh of December, we by Matt Wallace,

to

work

the prob-

an accelerated pace. Spear and Project Scientist Matt Golombek

were taking

a

keen

interest in

our progress against the noise. They began

holding weekly meetings with

Economou and

representatives of the

rover team.

Wallace

knew

that

whenever the APXS was operated by the

produced noisy

spectra.

benchtop, with

its

from the

rover,

it

own

would

He

also

knew

that

if

he

set

rover,

it

up the APXS on

a

independent power supply and deliver clean spectra. His plan

totally isolated

was

to start out in

the bench configuration, and then very slowly, step by step, modify

it

He would take spectra all along the noise. He would do his experiments on

toward the in-the-rover configuration. way, waiting for the

first

sign of

the Marie Curie rover, protecting Sojourner dling.

With Economou

from any unnecessary han-

present, the search for the noise

began anew.

SOJOURNER

180

Wallace's systematic, deliberate approach quickly achieved

some

suc-

APXS cables passing too close to the rover power supplies on their way to the APXS electronics. The signals from the APXS sensor head were carried to the APXS electronics box by four coaxial cables. Due to the rover's thermal design, the cables were forced to cess.

The

tiger

team traced the noise

follow a circuitous route.

to

The sensor head sat on the back of the rover, while

the opening of the cable tunnel "igloo," the only

way into

the

WEB, was

at

the front of the vehicle. So the cables were routed along the side of the rover, across the front, through the labyrinthine path of the cable tunnel, nally into the

made

WEB

proper to connect with the

APXS

for a long set of cables; the longer the cable, the

and the bigger the chance of picking up

The

and

fi-

electronics. This

weaker the

signal,

stray radiated noise.

rover engineers rerouted the coax cables inside the rover as far

away from the power

To further protect the

supplies as possible.

cables,

they were then wrapped in copper shielding.

With the hardware modifications

on the Marie Curie

in place

rover,

He declared the instrument performance "adequate" to fly. He hoped that we would make further improvements. Economou would not be pinned down. He refused to reveal any quantiEconomou viewed

measure of what would be "good enough." Doing so would allow

tative

the rover

work on

team

the

instrument:

produce.

We had

the spectra.

to declare victory. But that

APXS. Economou preferred

to keep the

The more time they spent on

And Economou wanted the best

it,

APXS

stop

team focused on

the better the data

it

his

might

data he could get.

declared victory anyway. Sojourner

a solution to the

solution

would mean they would

was ready

noise problem running

for delivery.

We

on Marie Curie. The

would be implemented on Sojourner in about six months, during

the already allotted rework period.

When

the rover

team returned from vacation

plan had changed. again,

and asked

The

project

for the

of the year, the

had delayed the delivery of Sojourner yet

Marie Curie noise

journer immediately, before

after the first

delivery.

The

fix

to be

rover

implemented on So-

team agreed.

The Noise That Wouldn't Die

The rover

181

rerouted the cables from the sensor head

flight technicians

and added the copper-braid shielding to Sojourner, duplicating exactly

what they had done on Marie Curie. So the

new

spectra collected in the

Sojourner configuration were a shock: The data plots

pronounced

noise.

Dismay was

thick in the

team thought they had the problem

The debugging of gan

first

the

to feel like a voyage,

team might never

still

showed

room. Until that moment, the

licked.

APXS resumed, going on for so

long that

it

be-

an odyssey in a strange land from which the rover

return.

Matt Wallace had volunteered to lead the newest incarnation of the tiger

team

that

was

power system help

him

investigating the

engineer.

APXS

More than

deal with the antagonistic

noise problem.

that,

He was

a skilled

he had an even temperament to

Tom Economou.

Prior to his career at

JPL, Wallace had been in the Navy, serving two tours of duty on nuclear

submarines. He'd been given that assignment partly due to his psychological profile.

He

could handle long confinements in close, claustrophobic

environments.

But working with

Economou's nature

Economou took

its

toll.

It

was

just not in

to consider the JPL engineers as partners in the effort

to solve the noise problem.

He

could see them only as adversaries, either

uninterested or actively seeking to thwart his desire for

good APXS

spec-

tral data.

Experience working with the Marie Curie, Sojourner, and laboratory

model APXS

APXS

units

electronics

bility to noise.

had taught the

had subtle

tiger

team

that the supposedly identical

differences that gave

This variation had

them varying

suscepti-

contributed to the team's frustration

during debugging: They would sometimes find an apparent solution using the laboratory unit, and then discover

Curie or Sojourner. In the

final

it

was

ineffective

on Marie

attempt to correct the noise problem

before the Sojourner rover delivery, the tiger team tried mixing and

matching electronics boards from

all

three

APXS

units.

combination of boards would produce clean spectra under

Perhaps some all

conditions.

SOJOURNER

182

After they put the hybrid together, Sirota

There was

still

some

was ready to declare

noise in the X-ray spectrum, but the alpha and pro-

ton patterns looked "quite good." The team would do a the

APXS

hybrid in the Copper

The Copper Room was

The

There was a

the

APXS were due

had done the

wrong with test.

possibility that the

to

unknown noise

earlier testing. If that

the

APXS

Rerunning the

inside

Why

power.

t

a Faraday

electrical noise

sources in the building where they

were

Copper

tests in the

then there might be nothing

so,

Room could prove that. spectra in the

The X-ray spectrum was

was somewhere

Economou was can

from any

and even

remaining noisy readings from

When Sirota and Wallace examined the

looked bad,

floor, walls, ceiling,

or the rover, but only with the conditions of the

they were as noisy as before. noise source really

evaluation of

The copper formed

which shielded whatever was placed

outside.

final

Room.

just that:

the door were lined with copper sheets. cage,

success.

inside the rover.

still

Copper Room,

The

corrupted.

When

the spectrum

very vocal. "The Russians gave us good

you?" Sirota described the moment:

"Economou

flew

it's our fault, that we have bad He walked right up to Economou's face

off the handle, accusing us, telling us

power. Matt had just had enough.

and started yelling

remember contact

.

.

him.

trying to pull

He

definitely lost

him back

a

it

little bit.

a

little bit

right there.

I

There was no physical

.

'After that

Sirota

at

week, Matt Wallace never worked on the

became the

liaison

between Economou and the

team. Apparently he was the only engineer

nomou for an

who

could

APXS

again."

rest of, the rover

work with Eco-

extended period of time without wanting to wring the

sci-

entist's neck.

Time was

up.

There was nothing

left

to

do except reassemble So-

journer and prepare the rover to be handed over to the lander. Further forts to solve the

remaining noise problem would be restricted to working

we were able to implement on would be installed on Sojourner when the flight rover was hands, a few weeks before we shipped it to Kennedy Space

with Marie Curie. Any improvements Marie Curie again in our

ef-

Center for launch.

The Noise That Wouldn't Die

Many is.

rover

team members wondered why we couldn't just

To them, the APXS was

was the rover

wanted

APXS

APXS

APXS

as

given Economou's combative attitude, nobodv

his or

way

her

to help

getting

him

out.

It

seemed

what he deserved

for

to the en-

being accu-

rover:

Once Sojourner was on

to fire the pyros

and explosively jettison

ejection system

Mars, we'd send the the

the

and demanding. Some of the engineers joked about

satory, secretive,

an

flv

add-on to the important thing, which

just an

Economou was just

gineers that

installing

And

itself.

go out of

to

183

command

sensor head, sending

on the

flying yards away, never to trouble the

it

rover again. Sirota'' s

attitude

APXS. To the

was

project,

it

different.

Pathfinder,

and the rover was flight

was

just an

asked

APXS

deliver}' system. Sirota

had

experiments before, and the science Principal Investiga-

had been king: The engineers had

scientists

recognized the importance of the

was one of the premier science instruments on

worked on tors

He

for.

to deal

with whatever the

Principal Investigators did not have such despotic

authority But Sirota was

still

quite comfortable

needs of the science instruments.

with anybody, even

jump

Pathfinder, as a "technology demonstration" mission,

The

a bit different.

to

And

accommodating

the

Sirota figured he could get along

Tom Economou.

m The Pathfinder project remained vitally interested in the quality of the

.APXS spectra. The troubleshooting effort went on. Sirota.

The weekly meetings with

these meetings. Sirota pressed for nition of

when

a

on the need

lented and gave

them

complishment." Sirota Sirota got his

own

to provide a concrete defi-

enough.'*'

until launch. Spear

Otherwise he'd be

and Golombek backed

measurable definition of success.

for a

coordinated by

Manager continued. During

Economou

spectrum would be "good

working the APXS problem Sirota

the Project

now

Economou

number. "Getting that number was

a

a

major

re-

ac-

said.

harsh message out of the meetings: 'The

APXS

SOJOURNER

184

noise problem

was

a

go /no-go

pression that Spear did not

APXS." But

now

Sirota

situation for the rover.

want

I

had the strong im-

to fly the rover without a functional

had to contend with

a shortage of rovers. So-

journer was tied up in integration with the lander. Marie Curie was over-

booked, with every rover subsystem team competing for

test

time on the

vehicle.

The

solution

was

to build a

new APXS

over from earlier environmental boards, a test

tests:

testbed out of assemblies

WEB, and the lab unit APXS. At first Sirota wasn't even sure

WEB

the testbed could be constructed, since the test

same

left

the qualification rover electronics

size as the others,

wasn't quite the

and therefore the electronics might not

team had integrated the

early March, the mechanical

fit.

But by

pieces. Testing

could begin once more. Sirota

worked

one of Economou's engineers. As they

closely with

ercised the testbed,

it

began to look

like the

ex-

team's prior efforts had not

been in vain. The capacitors they had installed in October had reduced the low-frequency noise in the power supply

lines.

ing of the cables the tiger team had tried in radiated noise. But there

"This afternoon,

was

we began

still

very close to those which that

it is

APXS

Sirota

to be a

I

be dealt with.

to block higher-frequency

we achieved started to come APXS GSE alone, indicating

components which

quite ready to declare success:

this discovery, if

the

are causing the [re-

spectral noise."

wasn t

be surprised

filter

results

got from the

the high-frequency noise

maining]

aged by

we

"The

shield-

December had eliminated

a third source of noise to

tuning the

noise," Sirota reported by email.

The rerouting and

"I

am

however, given the history of

certainly encour-

this

thing

I

wouldn't

am overly premature and what am telling you turns out I

bunch of bologna." But when the

filter

was added

to the Marie

Curie rover, the spectra were clean.

They had beaten Economou's number. prove that the noise flight rover

was

for

filtering

would work

now impossible:

ing environmental testing.

It

Now

all

they had to do was

for Sojourner. Getting to the

was mated

to the lander, undergo-

Months would pass before Sojourner could be

pulled out of integration testing for

its

own APXS

fix.

SIXTEEN

SOUL OF SOJOURNER

What

the

hell's

the rover doing?

It

won't communicate.

It's

not

lis-

Do we have a hardware problem?" was exasperated. A hardware problem on Marie

tening to us. What's going on?

Henry Stone

Curie would be scary, because

it

might mean

a similar failure

was lurking

on Sojourner. Stone and Tarn Nguyen had pulled

any

ideas.

me into

the testbed to see

They had been doing a standard "healthcheck" of the

if

I

had

rover, the

kind we would do after launch. They were proving that the whole process of sending

command

sequences through the lander to the rover was

working.

the

And it was. When the rover woke up, it grabbed the commands from lander, operated the APXS to show that it too was still healthy, and

sent back telemetry.

supposed

Then Marie Curie put

itself to sleep, just like

it

was

to.

Stone and Nguyen

woke

the rover

up again

to give

commands

manually. But by the time they were ready to

what

it

to do,

had stopped asking

"What mission phase

is it

for

in?"

it

some more

tell

the rover

commands.

was

my

first

question. That

much

in-

formation was available from the debug port before the rover stopped talking.

SOJOURNER

186

Stone answered. "Phase "It

thinks

it's

on Mars.

2.

It's

What difference running

its

does that make?"

contingency mission."

# The rover operated in several mission phases.

way of drive

The mission phases were our wrong thing (like trying to

protecting the rover from doing the

around

and, even

if

inside the lander if

it

was

still

in space

on the way

the telecommunications subsystem failed, to

to Mars),

make

sure the

rover did something right. There were distinct mission phases active rover before launch, while in transit through space,

der before

it

stood up, and after

it

had

on the

when sitting on the lan-

rolled onto the Martian surface.

During normal mission operations, the rover team would send com-

mands

to

tell

the rover which mission phase to switch

to.

In addition, the

its own set of programmed rules that it checked every time woke up. If the right conditions were met, Sojourner would take it upon itself to progress from the current phase to the next. So if the lander

rover had it

powered-on Sojourner during

its

long voyage to Mars, the rover's on-

board sensors would detect zero-gravity instead of one Earth proceed directly to the "cruise" phase. time after landing,

it

and

When Sojourner woke up the first

would find itself in

38 percent of Earth gravity), and

gravity,

fair-to-middling gravity (actually,

move on

to

its first

"on-Mars" mission

phase.

And

for

each of the phases there was a separate "contingency

se-

commands from tell it what to do home stopped arriving. Suppose the rover inadvertently woke up in cruise, and the lander had no commands to give it. The contingency sequence would kick in. The rover would still do the right thing: go back to sleep, conserving its batteries. The on-Mars contingency sequences were designed to make the rover do everything from standing up, to driving down the ramp, to blindly trying to find rocks for APXS measurements. quence" already built into the rover to

if

# wrong with Marie Curie in the testbed? Nothing. The robot had performed exactly as it had been designed to The rover had started out in phase 0, or "prelaunch." When Stone and

So what was

do.

Soul of Sojourner

Nguyen ran their first test, they sent the check sequence that

months from now,

we

after

18?

standard, prewritten cruise health-

on the

already had

Sojourner was in space. They queued up the

quence on the testbed lander. The lander waited mit the sequence

as

ready to use several

shelf,

soon

as

dutifully,

Marie Curie requested

it.

se-

ready to trans-

Stone powered-up

the rover, and a few seconds later the lander and rover were talking to

each other. The rover pulled over the healthcheck sequence.

march through a

the

commands. Step

healthcheck command. Step

APXS. Step

4:

3:

1:

It

started to

Switch to cruise phase. Step

Send

2:

Do

few sample commands to the

a

Shut down.

The second time they woke up

the rover, Marie Curie

was already

in

much more

the cruise phase. But the onboard accelerometers indicated

than the zero gravity the rover expected during cruise. There was only

one place the rover was going ity

it

Curie for

was seeing must mean

jumped

after cruise,

that

it

and that was Mars. So the grav-

had reached

into the on-Mars phase, phase

commands,

there

2.

its

destination. Marie

When it

asked the lander

were none; Nguyen hadn't yet loaded the new com-

mands. With no

new commands,

gency sequence.

On

the rover began executing

its

contin-

Mars, a possible cause of faulty communication

between the lander and rover would be so the contingency sequence

was too

that the rover's radio

commanded

the rover to turn

cold,

on the radio

heater for ten minutes before trying to contact the lander again. Stone, Nguyen, and

lence ended.

The

I

waited a couple more minutes. Marie Curie's

si-

testing resumed.

# Jack Morrison was Sojourner's software "architect."

thousand software engineer, seemingly able to do merely

human programmer might

He was in a

a one-in-a-

week what

accomplish in a month. Morrison,

a

to-

gether with Tarn Nguyen, formed the entire rover onboard software

team.

Morrison liked to be not highest on his wife

on

left

alone.

He put it this way:

my list." He had worked at JPL in the

moved to Colorado where

thirteen acres of land.

When his

"Social activities are

'eighties,

then he and

they lived in a house in the mountains

employer began having

financial

diffi-

SOJOURNER

188

culties a

few years

later,

Jack and his family returned to Southern Cali-

Wilcox eagerly hired Jack back into the Robotic Vehicles

fornia. Brian

group.

No one knew much about him. He was always affable when approached. But he never said much about his personal life. He never joined the rest of the group in After his return, Morrison

was just

as private as before.

the cafeteria for lunch, preferring a microwaved meal in his office.

guessed that he viewed the lunch hour

as "quality

I

more work

time" to get

done, with fewer interruptions than usual since most everyone else was

out of the building. Morrison's principles for the rover software: "Early on

looking ture,

at

what

laid

I

I

It

had

down some

we had on some

to

be

the

philosophies about

as simple as possible,

CPU

visibility into

and

we

didn't

how

started

thought It

it

ought to be

had to be very

reli-

mainly because of the constraints

And we wanted

power.

on, since

know

it

a limited

was going

to

to have

be an engi-

everything about the environ-

ment it was going to be operating in. Looking at

embedded system, with

I

would be.

it

memory and

what was going

neering experiment, and

typical

I

thought the requirements were in an overall architec-

done, what the main goals of designing able.

when

that,

I

approached

it

as a

computing environment, always

how much memory something s going to take, how efficiently something's going to work, and how simple and straightforward having an eye on

you can implement

it

and have

it

do what you want."

Morrison's philosophy was perfectly in line with the rover team's desire to

keep the overall rover design

plified

by the choice of the 80C85 microprocessor

The computer

chip

was

as simple as possible. This

far less capable

of the time. But that simple

CPU

was exem-

as Sojourner's brain.

than the average

home computer

chip controlled a suite of sensors and

motors that would be the envy of any personal computer owner.

The its

was

rover's apparent simplicity

creators

nation of

by doing exactly what

all

it

of the design choices

had resulted in

a final

system

deceptive:

It

would often

surprise

had been designed to do. The combi-

made by

that, at first,

no

all

of the involved engineers

single designer fully grasped.

This complexity gave the rover something like "personality." While the

Soul of Sojourner

189

team members were learning the nuances of the system,

it

sometimes

ex-

hibited behaviors that confounded people.

many other computer-based machines, the rover only did one thing at a time. "We can't walk and chew gum at the same time" was how Unlike

Henry Stone described board, then

mand was

it

Over

in presentations.

its

radio link, the rover

new command sequence from the lander and store it onstart carrying out the commands one by one. Once a com-

would accept

a

would

finished, the rover

store the results into

telemetry packets, and transmit the packets to the lander.

asked the lander for a that

new

Then

the rover

new one had arrived now executing. If there was no

sequence, just in case a

might override the one the rover was

new sequence

one or more

—there usually wasn't—the rover moved on to the next

command.

Sojourner had

first

been brought to

life

months later, with only minimal func-

livered to the lander less than three tional testing.

The

The rover had been de-

in late 1995.

original schedule

showed more

time, but the needs of

the hardware had taken precedence. Centrifuge and acoustic tests at

Wyle

Labs, modifications to the rover brain, and the seemingly unending struggle to track

journer's

down and

first

fix

the

few months of

APXS

noise problem fully

consumed

existence. Functional test time

So-

dwindled to

only a couple of weeks. In his low-key way, Morrison

ing his software on the vehicle. that the software guys

I

warned

if

later.

They had hardware

functional testing, they delivered

On May thing else

17, 1996,

would take

On that day, months, the

rest

test-

of the core team

needed more time. The hardware guys nodded and

the hardware wasn't ready?

could happen

needed more time

complained to the

kept tight reign over the vehicle. After software

that he

all,

what was the point of

They

testing

figured the software testing

to deliver.

At the expense of

lost

it.

we would be reminded that the assertion that everycare of itself

was merely wishful

thinking.

Sojourner came out of hibernation. During the past few

flight rover

had been

tied

down

to the lander petal, exactly as

SOJOURNER

190

it

would be during its voyage

its

The lander had been

to Mars.

pyramidal form with Sojourner

had been subjected to various

ment during the seven-month

inside. Together, the

tests that

into

spacecraft

cruise phase.

During the it

entire experience,

would throughout the

real

come.

Now again at ings.

two

up

simulated their shared environ-

Sojourner had remained dormant, just as trip to

closed

A

the lander sat

last.

on

a platform in the cleanroom,

its

petals

open

A few bunny-suited engineers stood watch over the proceed-

of tables had been pushed together in a cluster in the next

set

room. Computer workstations were arranged on the ductor and the rest of the

test

The

tables.

test con-

team worked at their keyboards or watched

the activities in the cleanroom via closed-circuit television. Step by step,

they marched the lander and rover through a simulation of their

first

day

on Mars.

The

NASA sponsor had flown in from Washington, D.C., to obHe and Jake Matijevic stood off to the side, watching and

rover's

serve the

test.

sometimes speaking back and forth

The

cable-cutter pyro fired

the table.

A

in

low

and the

tones.

single rover

few of the engineers applauded. More pyros

the lander's grip

on the rover

Those of us

itself.

watched the video monitor. Sojourner began to wheels drove forward, while the

by

their restraining hooks.

soon!

The

The

rest stayed

The telemetry transmitted from

know

it

had

a problem.

few minutes, the rover would

in the test control rise off

unmoving,

the petal.

Its

rear

It

stopped too

failed.

showed no anomalies. The

The sequence was

try driving out of

After a quick consultation over the voicenet,

room

held in place

still

The stand-up had

the rover

across

fired, releasing

rover rose. Almost. Almost.

rockers dropped back down.

rover didn't

ramp unrolled

we

its

still

running. In a

remaining

restraints.

decided to complete the

stand-up manually before that happened. In the cleanroom, a white-

smocked rover mechanical engineer springs between the

telemetry

still

There was

lifted

articulated halves

the rockers until the bent

snapped into

place.

The

rover's

showed no problems. a hint of

—

movement

Then the rover stopped dead. rover reporting an error.

a barely perceptible roll

backwards.

Now what? The telemetry came in, with the

Tarn Nguyen looked up the obscure error mes-

191

Soul of Sojourner

sage,

which led us to the APXS Deployment Mechanism. For some

APXS was

son, the rover thought the

budge. That

made some

When

ployment mechanism. the

ADM

sense:

deployed, so

was refusing

it

We had designed in a fail-safe

APXS sensor head into the dirt, leaving time. What would happen if we forgot to

to put the

there for hours at a

to

for the de-

we would

the rover got to Mars,

rea-

be using

the sensor retract the

ADM before driving on to the next target? You didn't want to accidentally drag the sensor head through the

APXS, or

at the least

cover

it

dirt. If

with

soil,

you

did,

you'd likely damage the

which would make

further experiments. So Jack Morrison had built a

Whenever the rover started to move,

software.

it

useless for

self- test

into the rover

the software

compared the

ADM potentiometer reading to a preset threshold value: If the new readwould conclude

ing was bigger than the threshold, the software

that the

ADM was deployed, and rover motion was disallowed. You could look at the rover and see that the ADM was stowed. rover thought otherwise.

ponder pot

is

if

we hoped to

bad,"

I

were broken:

We

had a mystery, one we

continue the

suggested.

The

just ignored the

bad

tentiometer broken,

the rover that the

test.

"Tarn,

let's tell

rover had a

way to

deal with sensors

used an alternative sensor,

it

if

there

would be

willing to

move

it

knew

was one, or more often

ADM po-

sensor. Perhaps if the rover considered the it

The

didn't have time to

again.

No such luck. When we tried it, we got the same error message again. I

was stumped. The lander

testing

went

on, but the rover

was done

for

the day.

Our NASA sponsor seemed

He

rover problems.

to be understanding in the face of the

expressed relief that the rover had survived

cruise

its

testing with the lander.

The

rover core

team meeting on the following Monday was

gloom, doom, and finger-pointing. Howard Eisen called the journer to stand up during the call."

test "a fiasco." Matijevic said

Suddenly, everyone in the rover

it

of

full

of So-

failure

was a "wakeup

team meeting had ideas for "fixing" the

problem with software and operations: There should be troubleshooting guides written

.

.

.

The software team should be more

personnel should be trained to understand the software

dedicated .

.

.

The

.

.

.

More

core team

should begin discussing operations issues during the weekly meetings

.

.

SOJOURNER

192

I

fumed.

I

had been warning the team

of skipping software testing. risk,

I

but the hardware comes

schedule was cut, so

much

first/'

so that

Henry Stone had warned me anger,

I

I

told,

had raised the

no one wanted

that the rest of the

And now most of the team was Even through my

for the past year of the hazards

had always been

"We understand

the

issue each time the

to hear

it

anymore.

team was tuning me

out.

surprised that there had been a problem?

feared that the

new attention to the issue would

not bring solution, but instead a set of naive quick fixes that would merely

compound

the

Later, Jack

wasn't a

initial error.

Morrison expressed the software engineers' reaction:

wakeup

tention to

call to us. It

what we were

was

wakeup

a

call to

them

"It

to start paying

at-

saying."

Within a couple of days, Morrison

had

analyzed the rover

fully

Rover stand-up was a tricky procedure. The rover drove

its

failures.

rear wheels

forward until sensors on the rocker-bogies indicated that the rover had stood up just enough to latch the "broken

wheels stopped.

If

in

normal

rockers in place.

on the wheels would

would be subjected

driving.

The

ting Sojourner into

its

start to get bent,

to stresses they

would never encounter

stood-up configuration several times to account for

Due

to their abbreviated test schedule

before Sojourner's delivery in January, Morrison and

stature.

from

calibration.

Nguyen had never

So the rover had just followed

a reclined position to

what

its

sensors said

its in-

was

full

Mission accomplished. Only the rockers hadn't locked into place,

so as soon as the wheels stopped driving, the rover sagged, as

weak

would

and the

Complete" could be determined only by put-

variations in the sensor readings.

completed the sensor

the

precise values of the rocker-bogie sensors that

translated into "Stand-Up

structions, rising

Then

the rear wheels kept driving past this point, they

start slipping, the cleats

rocker-bogies

'

in the knees. This

the rockers

was

if it

were

exactly the failure that contact switches

would have prevented. Unfortunately,

it

was too

late to

on

add

the switches into the design now.

The

practical solution to the problem: Calibrate the sensors.

And

Soul of Sojourner

maybe improve

that algorithm for standing

thing did go wrong, the rover would detect

193

up the

it

rover, so that if

some-

and skip any future motions,

thus avoiding even worse trouble.

Nguyen and I had almost solved the second failure utes during the

test.

We

APXS deployment mechanism. The ADM posifail-safe threshold. But we hadn't known that,

was indeed beyond the

even with the

member the

min-

had correctly interpreted the error message and

traced the problem to the tion

in those tense

ADM potentiometer tagged as

the last

known

position of the

most recent information

trusted.

it

"bad," the rover

ADM,

We

months

before:

The number was

would have worked. time to review

all

re-

and refuse to move given

learned from Morrison that

what we should have done was change the threshold

The threshold value had been improperly

would

so small that

In the rush to deliver the

number.

to a better

an impossible value

set to

five

ADM

no position of the

FUR, there had been no

the values and catch the error.

Any attempt

to drive

Sojourner after the parameter had been set would have revealed the

problem. Ironically,

none of these problems were with the rover software

Given the calibrations and parameters

had done

perfectly everything

had not been

it

it

itself.

had been provided, Sojourner

had been asked to

do.

Only the

results

perfect.

w The acrimony within the rover team went on for about two weeks. it

was evidenced in email

When

my

I

traffic

calmed down,

concerns,

how we had

I

Much of

among the team members.

composed

a

long email message describing

gotten into the current situation, and what

I

thought we needed to do to get out of it. Primarily I was making a plea for

more

test time.

The philosophy in the

testing until "later";

I

warned

past

that "later

is

had been to postpone software now."

Eisen responded with a scathing email telling software team was doing tioned the his

commitment

me

wrong and should now do

all

the things the

differently.

He

ques-

of the software engineers, while reaffirming that

mechanical team was "prepared to do whatever

it

takes."

SOJOURNER

194

I

tribe,

quickly banged out

then rewrote

it

my answer to what seemed to me a wanton diait read as if my temper were

again and again until

under control.

The email volleys died down. The team moved on to other immediate issues.

After

the

all

Sturm und Drang,

little

changed. Morrison and Nguyen

did create an on-line troubleshooting guide to explain error messages.

Marie Curie and Sojourner remained overcommitted, their presence

re-

quired in system tests together with the lander. As Matijevic viewed there

was "precious

little"

time available even for these

nally scheduled for the software

performed were

still

team

to test the rover

tests.

Days

and learn

that the

APXS

noise

Marie Curie, be implemented on Sojourner

fix,

even

if it

at least a little

for testing

early,

instead of waiting until

a success. Perhaps there

time for testing with the lander, with a

For the next few months

APXS was

bit

would now

more

left

over

it

seemed

to the software

team

that they

either participating in lander system tests, or getting ready

for the next one.

erating

on

on our own.

were always

these tests

it

meant pulling Sojourner out of project system tests.

The APXS surgery on Sojourner was be

how

already demonstrated

the scheduled time in July Spear wanted proof that the flight fly,

origi-

often lost to dealing with hardware issues.

Tony Spear insisted

ready to

it,

There was nothing

ideal

about the conditions, but in

we were getting to know the rover.

Slowly,

all

we got better at op-

it.

And the a beast our

rest

little

of the rover team began to appreciate just

Sojourner really was.

how complex

SEVENTEEN

LICENSE TO DRIVE

the start of MFEX, Donna Shirley was extremely conscious of From her limited budget. The cost-capped $25 million for the rover had to

cover everything: not just design, parts, development, assembly, and tests,

but

all

operations during the mission as well. Shirley's cost concerns

led to her decision to rely

on commercial radios for rover communication.

She also hoped that the rover control station already developed for the

Mars Science Microrover demonstration would be good enough

manding the new But the

for

com-

flight rover.

flight rover

would not be

like the

Mars Science Microrover.

Sojourner would respond to different commands, and there would be

many more of them, for operating its cameras, instruments, heaters, and behaviors. And every control station we had ever created had sent only one command at a time: You told the rover to do something, you waited to see the results, then

had been just

told

fine for research.

would only get

we had

you

a

it

to

do something

But when the

chance to send

new

else.

That approach

rover got to Mars,

commands about once

we

per Martian day.

make those chances count. Instead of sending individual commands, we would send a sequence of many commands maybe hundreds of commands all at once, and all in the proper order, which would

So

to

—

—

tell

the rover

what

to

do over the next twenty-four hours or more.

SOJOURNER

196

MFEX

Nearly a year after the nally relented: "Bring Brian

team had come

into being, Shirley

fi-

Cooper onboard." Stone and I had convinced

her that cobbling together a few improvements to the Mars Science Mi-

would not enable us

crorover control station

to operate the Pathfinder

on Mars. Cooper had worked on the control

rover

rover since

JPL the days of the Blue Rover. The true complexity of com-

manding the

flight rover

would be revealed during

In the end, only about 10 percent of the rover's

station for every

MFEX

development.

commands would have

anything to do with telling the rover where to go, with the rest associated

with the rover's onboard instruments and with maintaining the health of this six- wheeled interplanetary spacecraft.

Brian Cooper

became

one-man subsystem, single-handedly

a

devel-

oping the software for Sojourner's Rover Control Workstation (RCW).

Now officially a member of the control and navigation team,

Cooper be-

gan attending Stone's weekly team meetings, but quickly got bored:

Henry

ever talks about

journer's

is

parts."

Cooper knew

components arrived on schedule was important, but

very relevant to designing the

Cooper did

care about:

We

RCW. There was one

parts

it

procurement

figured

it

Inc.

the software that

RCW

took a long time for Cooper to believe that Pathfinder was

It

wasn't

ordered a newer, faster Silicon Graphics

computer upon which he would create and run

(SGI)

would be the

"All

that ensuring that So-

would get

cancelled."

Or maybe

Pathfinder

would

fly,

real. "I

but the

rover wouldn't be onboard. People at JPL were always talking about rover

missions to Mars.

been est

a big deal.

He remembered when Mars

Then it had gone away

Rover Sample Return had

Now they were saying that the lat-

microrover was going to Mars, and he was part of that mission, de-

signing the newest vehicle control station based

learned from years at JPL.

He

all

He

the workstations he had didn't

want

on everything he had

worked on

to set himself

up

for the past eight

for a big disappointment.

wouldn't permit himself to become too emotionally involved in the

mission. He'd go celled

it

get cut.

home

at night

and

talk to his wife:

"They haven't can-

yet. We'll see." But Pathfinder didn't disappear.

The

test

model rover evolved

into an ever

more

Its

budget didn't

flightlike config-

License to Drive

uration. And, in time, parts of the

19?

SIM and

the

FUR began

to

come

into

existence.

Cooper began

to allow himself to believe. "We're really going to

He opened the door partway to view the enormity of what he was a part of. Like many of us on the team, Cooper had grown up reading science fiction. He and I had often discussed Mars. We're really going to do

the reason plied

working

more than

at

this."

JPL was

"We

ever before:

As Henry Stone's system engineer, sion the rover to do.

it

at first,

I

I

excited

but

to

It

just

seemed easy

when

to

the time came.

now had in his hands. Cooper and

talks

about what the Rover Control Workstation should look like.

the software designer and developer;

Together

we

fell

I

I

new

picture,

move

RCW's

soil to

study

its

again.

new site, turn around, deploy the APXS against a rock, APXS data, then shut down for the night. Overnight the

continue gathering data; the rover would wake up for a few

When morning finally

ask the lander for whatever

of

commands had

activities

would

take

dred commands. For those of us operating the rover,

new computer program

had to

hit

the rover

every day.

And

was supposed

to be.

its

it

its

IMP camera and took

Those

then shut

pictures

recently arrived

two or three hun-

would be

like writ-

like a stage entertainer,

our marks on time: The rover had to reach

day before the lander aimed

far,

came, the rover would wake up and

new batch

from Earth. Orchestrating these

ing a

drive to

properties, take a

minutes every few hours to read out the data collected so

down

and

design.

to a

start collecting

APXS would

one wheel into the

He was

into the role of customer

defined the requirements for the

location, dig

soft-

started having long

During an "average" day on Mars, the rover would wake up, a

know

by the power of the new SGI hardware and

tools that

user.

tell it

my perspective was already that of the

ware

he

ap-

was constantly thinking about the mis-

was going to need from the control station

And Cooper was

it

get to turn science fiction into reality"

eventual operator of the completed rover.

what

Sojourner took form,

as

would perform on Mars and what we would need

didn't realize

I

and

special,

had

we

destination for the pictures of

to be sent

where

back to Earth

SOJOURNER

198

before the day's communications

we

them,

wouldn't

window

closed. If the rover wasn't in

know enough to plan the

next day's driving.

For planning the rover's traverses through the terrain, we'd put on stereo goggles like those we'd used for the

switch to the

Mars Science Microrover, and

RC W's CARD display. Within this display window, we could

view the terrain

Because the

in three dimensions.

IMP camera on

the

—only about fourteen degrees

lander had such a small field of view

looking It

at just

one image

at a

your head

CARD window was

ages arranged three across by rover in the terrain

really six

two down,

was represented by

.

all .

.

windows:

The

in stereo.

the rover.

six

IMP

im-

position of the

Cooper had created

3-D icon of Sojourner that could be turned or positioned anywhere

the scene; even it

blinders and not being able to turn

or right. You'd always be wondering about what you

left

couldn't see. So the

a

time just wasn't going to be good enough.

would have been like wearing horse

its

in

would change, depending on how close or far away

size

had been placed.

Anyone who uses allows

him or her

up-or-down

to

a personal

move

we wanted

Spaceball

two dimensions

a pointer in

that

left-or-right

and

—

The

display.

RCW

in three dimensions, as well as specify the direc-

the rover to face.

was the next

mouse

familiar with the

—to reach any location on the computer

would control the rover tion

computer is

step

We

needed

beyond the mouse.

a

new It

input device.

looked

The

like a baseball

mounted on a post. The post rose out of a base that was bolted to the desk next to the keyboard.

If

you wrapped your hand around the Spaceball and

pushed, the 3-D rover icon would tance, getting smaller If

you

and

pulled, the rover

suddenly disappearing cally

fly directly

smaller, until

away from you

you couldn't even

came racing back, quickly

when

it

filling

into the dis-

see

it

the screen, then

reached a position that was mathemati-

behind you. Push a button, and the rover would return to

position, visible again. slid left

really

Push

or right in the scene

move:

It

its

"home"

or right on the Spaceball, and the rover

on the

screen.

The

Spaceball itself didn't

sensed the forces you were exerting on

and the rover turned, or rover

left

anymore.

rolled, or pitched forward.

it.

Try to twist

When

it,

you had the

model positioned where you wanted it, you pressed another button

199

License to Drive

to lock in a waypoint, the place

humor led him

sense of

show

the soil to

We I'd

the user

you wanted the

where those waypoints were

say something like

Sometimes

"I really

window

RCW evolved.

need to know what time they're going to happen.

I

able to schedule an experiment at that?"

in the

in

need to know how long commands will

need to know that time

I'll

Cooper's

3-D images of lawn darts embedded

to design

continued our brainstorming sessions. The

take to complete.

do

real rover to go.

Cooper would add

in

Mars time. That way

Mars noon

to his

list,

if

we'll

we want to. How

and new features would

can

be

we

time

in

magically appear in the software.

The

first

test

schedule

—

of our

as

ability to create rover

command

we would face with the rover on Mars

sequences on a tight

—would come during

early rover environmental testing.

Just as

we

subjected the Marie Curie to high accelerations and vibra-

tion to ensure that

voyage to Mars, that the rover

we

it

would

survive the launch and landing loads of the

also stressed

it

with "thermal /vacuum"

tests to

prove

would handle the temperature extremes of the Martian sur-

face without failing.

The thermal and vacuum

tests

were combined be-

cause heat flows through a spacecraft differently in the

vacuum of

space

than in the atmosphere of Earth.

The Marie Curie thermal/ vacuum test was scheduled to begin on October 22, 1995.

The plan was

to take the rover through five Martian days

of operation. Al Wen, the thermal engineer on the rover team, already

had

a

mathematical thermal model of the rover, one that predicted

how

heat would flow as various devices were turned on and off and the temperature outside the rover rose and

fell.

But

how good was

the model?

How fast would the modem heat up once its heater was activated? Much of the energy used by the devices onboard would end up as waste heat.

(The WEB's design counted on the

This heat would

warm

the interior of

WEB enough during the day so that its contents wouldn't freeze dur-

ing the night.)

too

it:

warm

Would the

as the

electronics inside the

Warm Electronics Box get

day progressed, forcing the rover to shut

down

early in

SOJOURNER

200

the afternoon?

The thermal /vacuum

needed to validate

his

test

would

give

Wen

And of course it would prove

model.

the data he

that the rover

could survive the extremes of the Mars temperatures. there were any surprises during the

If

would

there

test,

still

be

a

chance to make modifications in Sojourner to correct them.

The plan for the test required that

The performance of pieces of

we

exercise the rover in a realistic way.

the rover might change with temperature.

Unexpected interactions between those pieces might further impact the

way tell

And what we

the rover acted.

what

the thermal guys

ing the

The

test.

They

felt

change for the next

to

Henry Stone and Allen

learned on one day of the test would

Sirota

and the team just weren't yet

that the rover

ready.

support team was pushing hard just to get the equipment ready

test

in time for the reserved dates the test ticular

day.

both had strong doubts about perform-

chamber was

was concerned over how short

available for

available.

Stone in par-

a period Marie Curie

performance testing before the thermal /vacuum

had been

test.

Those

of us on the software and operations team were only just beginning to learn

bugs

how

the rover

worked

as a system.

There were plenty of software

to uncover. Stone argued with Matijevic that if

left

wrong with

the rover while

it

something went

was in the thermal chamber,

we would not

be able to determine whether the problem was a consequence of the

mal environment we had imposed on the lated

rover, or

bug that happened to crop up during the

test.

due to

having the thermal /vacuum

pone the

test for a

Matijevic

ment

month

a totally unre-

Such bugs might mask

actual thermal limitations. So if the team's experience sufficient to distinguish these types

ther-

was currently

test at all?

Perhaps

it

would be wise

to post-

or two.

would have none of

it.

He

did not disagree with the assess-

that so far the team's overall experience operating Marie Curie

disquietingly small. But

opment

schedule, he

in-

of anomalies, what was the point in

when he looked

at the rest

saw no future opportunities

was

of the rover devel-

to repeat the thermal/

vacuum test. The schedule was too tight. He was certain that there would be no second chance. The test would take place now.

201

License to Drive

Each day of the

test,

next to the chamber. far,

there

would be an afternoon meeting

The team would consider the progress of the

and identify changes

cuss options. Eventually,

would

I

would have

I

Eisen, Matijevic,

around the big

sit

test so

table

and

dis-

we would go through the "strawman" command

sequence, marking changes. changes, since

Layman,

in the next day's plan.

the thermal engineer, and others

room

in the

had

insisted

on

a 3 p.m. deadline for those

commands, document

to write the final

them, and generate the coded versions using the Rover Control Workstation.

But the

test

team often missed the

ing knowing that each minute of delay the early hours of the

command

necessary

need to come

in to

morning before

deadline.

would

sit

now would push me I

meet-

in the

further into

would be done producing the

sequence for the next

work

I

day.

And

of course

I

would

early that day to ensure that the sequence

loaded properly and that the

test

was

could proceed. So as the planning meet-

my temper rose higher and higher, often evidenced by my the others at the table. When tomorrow's plan was finally

ing dragged on,

snapping agreed

at

to,

I

ran out the door and across most of JPL to Building 107.

At the moment, the control station was to support these tests:

was

still

a prototype, barely ready

Almost every command the rover could execute

available via the control station, but there

was no mechanism

printing out the results in a human-readable form. to

go back and forth between the

was using

And

a result,

since the

RCW was its

still

in

development,

software as

I

stances, the thermal /vacuum test

commands had More than once that neither

it

first

was forced

team could

wasn't unusual for

was entering sequences. was the

I

spreadsheet program

to generate the reports that the rest of the

uncover bugs in

In

the

failing

tried out before,

of the

home and we would try to

me

to in-

time the particular rover

RCW software crashed when

Cooper nor I had

I

read.

some

ever been used in a sequence generated by the

encountered such a at

RCW and the

As

for

I

and

tried to I

RCW late at night,

had I

do something

to start over. If

would

reason out a work-around.

RCW.

call

I

Cooper

SOJOURNER

202

We found no single probWe had proven that de-

Marie Curie survived the thermal /vacuum test.

lem

traceable to the rover's fundamental design:

sign sound.

Yet the testing had gone anything but smoothly. Sirota logged twenty-

two new problems. There were problems with the rover's

onboard software, causing certain

skipped. There

early version of the

commands

to be unexpectedly

were problems with waking up the

rover stand up, taking

APXS

municating, and even

some minor hardware

spectra, operating the

if

we had had

WEB

heaters,

com-

errors in the electronics

Many

boards that had not been caught previously.

wouldn't have happened

making the

rover,

of the problems

time to learn the proper steps to

operating the rover. And, as a self-taught "student driver,"

I

had put

a

few

my command sequences.

errors into

As Stone had predicted, nearly revealed in functional testing,

vacuum test.

Fortunately,

all

if

all

there

of the problems would have been

had been time before the thermal/

of these problems could be corrected within

the remaining rover integration and test schedule. Given the rover team's fears

going into the

test,

I

rated the results as

Not Bad

at All.

Matt Golombek, the Project Scientist, had organized a field trip for

the Pathfinder scientists and engineers.

group

to the

The

State, a region

Golombek had determined was likely to echo

that

Ares

Vallis,

ago or

the chosen Pathfinder landing

so, a catastrophic

would introduce the

trip

Channeled Scabland of Washington

some of

on Earth

the geologic history of

Thirteen thousand years

site.

flood had formed the Scabland, creating channels

and carrying boulders and smaller rocks great strewn around the region. The purpose of the

distances, leaving

trip

was

to give the

them team

insight into geological formations the flight lander and rover might en-

counter during their mission. As a future rover operator, Brian Cooper participated.

During the Scabland

trip the project also

made one of

its first at-

tempts at outreach to increase the public's awareness of the mission, inviting about a els.

When

dozen educators

to join the Pathfinder

the buses stopped at each

site

team

by

in their trav-

within the Scabland, the whole

License to Drive

203

group would coalesce around Golombek while he described terrain features they

were

seeing, their geology,

them

to be. But other than including

had been grafted onto the

Cooper felt more of of the

scientists: "If

And

things."

unlike

I

wasn t an

do with the group of teachers

I'd

be a teacher.

that trip

in Everett,

Washington.

of the other engineers, Cooper wanted to talk find out their reactions to Pathfinder. "It

When

how operations would be

By

Cooper would

filled

with questions

done.

the conclusion of the field

speak to her elementary school

trip,

class

Cooper immediately agreed:

rover."

teacher

that

O'Rourke discovered

be sending commands to the rover on Mars, she was about

many

love to explain

I

met Fran O'Rourke," an elementary school

was during

I

most of the

with the teacher group than with

engineer,

group of educators, and

to the

to

come

trip.

a kinship

many

they had

in these discussions,

what

Pathfinder team wasn't quite sure that

how

and

in detail the

It

O'Rourke had

invited

Cooper

to

about Pathfinder and "driving the

sounded

like fun,

and he

liked the

idea of getting kids excited about science and space.

What Cooper

didn't

know was

that

O'Rourke had bigger plans

him. O'Rourke schemed with Cooper's wife, of the rover team.

for

who then enlisted members

We made sure Cooper had a rover to drive when he got

to Everett. "I

brought Go-For

2,

one of our prototype

rovers."

Go-For 2 was

simplified descendant of Brian Wilcox's fork-wheeled Go-For. "It tually a purely teleoperated vehicle,

system.

It

could be

was great

like.

So

Cooper had

I

for

demos.

It

had

a

was great

camera and for

was

a

ac-

a video broadcast

showing kids what

a rover

brought that up."

anticipated that he might

end up talking

to

more than

when he walked into the auditorium of Cedar Wood Elementary School, the room was completely filled with people. Every class in the school was present. And many of the parents of one classroom

full

of students. But

those students were there as well, lining the walls. "They invited the press. It

was

a big deal for

them. They had actually invited

a lot

of people from

neighboring communities." First

how

Cooper spoke

to the crowd, describing Pathfinder, the rover,

and

he would operate the rover on Mars. "They had an incredibly smart

SOJOURNER

204

group of kids there asking great questions. The students created and maybe hundreds of things from it

would be

with

like to drive

first

on Mars," and

his seat belt on, as if

Then came Martian

telling

made

to

be

safe,

and to drive

me the world's

the challenge. "They were going to award I

passed the driving

and the governor of Washington to

ton State trooper oversaw Cooper's "driver's

Cooper

him

posters,

me what

different grades, asking

he himself were going to Mars.

driver's license, if

representatives

all

test."

They had

test.

officiate."

The trooper

to put the rover through a series of maneuvers.

Mars

these papier-mache

terrains

and had

state

A Washingdirected

"The kids had

me

drive

Go-For

around." "Fortunately

The

I

passed,"

students had

cense, with

many

Cooper commented with mock relief.

competed over the appearance of the

of them submitting designs.

"It

was

driver's

li-

way to get The governor

a great

the school kids excited about our mission and space travel."

presented Cooper with his Mars driver's license, far too large to carry in a wallet, but suitable for framing.

was

In

Another copy of the Mars

driver's license

sent to the Smithsonian collection in Washington, D.C.

time

we were using the Rover Control Workstation

Rocky 4 rover

in the Building 107 sandbox.

lander, sitting in

how well the

Donna

dart waypoints; then

nagged the rover team about

the rover. She didn't trust the results

course

MSM

the

simulated

we watched

rover did actually getting to the target.

Shirley periodically

Marie Curie,

leftover

command

one corner, took the place of the IMP camera. Cooper

donned his goggles and designated lawn to see

The

to

in the sandbox.

we

got running Rocky

field testing 4,

and

later

Even driving through the new outdoor

test

—dubbed the "MarsYard" and painstakingly created by the rover —was not enough. Shirley would be placated only by

technology program a true field test,

somewhere out

with Marie Curie navigating through natural terrain

in the desert.

She also pointed out that

Hank Moore,

the geologist

who had been

chosen as the Rover Scientist for the Pathfinder mission, thought stereo images could be misleading. He'd been part of the Viking team

many

License to Drive

205

years before, and had studied stereo pictures sent back

landers of that mission.

Moore wasn't

mating where the rover was: images had

been

all

concerns, and

It

sure

from Mars by the

you could do

seemed to him

good job

esti-

that the rocks in the Viking

closer than they appeared. Shirley

wondered when we would

a

echoed Moore's

finally take the rover

out to a

"real" location.

When

I

heard of Moore's warning,

first

Moore might have

I

bristled.

Hank

figured

I

credentials as a scientist, but not as a rover operator.

Estimating locations precisely in stereo images was exactly what

we had

CARD to do. We had already demonstrated that it worked times. Who was this guy? At the time, "Hank Moore" was just a

implemented several

name

to

me;

I

didn't

know that getting to know the man would be one

the great privileges of being Field testing

was going

on the Pathfinder mission.

to be a logistical headache.

uncovering subtle problems even with our sandbox the desert before

we were

ready? Shirley prodded.

change seemed to occur every few months. promising that the

field test

It

I

And we were

tests.

Why go

out to

usually ended with

"Just so

still

protested. This inter-

would happen when the time was

something like

Shirley acquiescing, saying

of

right,

me and

long as you get around

to

it."

In

September 1996, three months before launch, Brian Cooper and two

Rover Control Workstations moved from Building 107 to the Space Flight Operations Center. By now, the all

of the bugs had been

slain.

RCW had become a solid system. Almost

There would

ditions to the software, because experience

builders to ask for

new features. But the

still

be tweaks and minor ad-

would lead the rover sequence

RCW was alreadv the workhorse

of our systems and operations testing, just as

it

would be when the

"tests"

were on Mars.

The second

floor of Space Flight Operations Center

of the Pathfinder operations team was already there. neers

who had worked on

in flux.

Most

of the engi-

the development of the lander's hardware and

software were moving out, fice cubicles

was

Some

making room

for those

who would

were going away, changing shape, re-forming

into

fly

it.

Of-

huge con-

SOJOURNER

206

ference

up

rooms and an open "bullpen" where

shop. Pathfinder's Mission Control

room,

the science teams called the

MSA,

MSA

identify

Cooper's door read simply

"ROVER CONTROL." The

eventually hold three or four

working furiously together into the right

cute

would

More

commands.

members of

office

it

and transform them

commands Sojourner would

All the

—

was big

the rover uplink team,

to plan rover activities

ever exe-

construction paper greetings and messages arrived from the stu-

Wood Elementary. Cooper tacked them up on the bulletin

in the hallway outside his office, so that

finder could see the excitement their mission

Jake Matijevic,

now the

everyone working Path-

had already engendered.

down the hall. and I had moved our offices

Rover Manager, had

Within a couple more months Henry Stone to just

and elsewhere

originate in this room.

dents of Cedar

board

floor.

key mission functions. The sign over

around the second floor to

would

set

for Mission

Support Area, was already in place in the center of the second Blue signs with white letters went up in the

would

his office

around the corner from the rover control room.

Brian Cooper surveyed

all

the

activity.

"We're really going to do

this!"

EIGHTEEN

METEORITES, E1EE, AND JOB SECURITY

On

August

D.C.,

7,

1996, scientists at a

announced

news conference

that possible evidence of

life

in

Washington,

on Mars had been

discovered in an Antarctic meteorite. Analysis of the rock had de-

termined that

it

was indeed of Martian

origin.

the meteorite, which had been designated state that they

past

life

The

science

were not claiming to have found proof of

on Mars. They only

team studying

ALH84001, was very

said that there

were

a

careful to

either existing or

number of

separate

pieces of evidence that could together be plausibly explained by a biological process,

and only somewhat

less plausibly

tion of chemical reactions. There

understood through a combina-

were even

tiny shapes, visible only in mi-

croscopic images, that the scientists proposed "might" be micro-fossils.

the ing.

The media wanted to know what the impact of the news would be on Pathfinder mission. The answer: none. The press found this perplexThe

possibility of

life

on Mars was

a big story

How

could the next

mission to the Red Planet ignore what had just been discovered?

To the engineers working on the Pathfinder over meteorite

ALH84001

planet they were targeting. it

would

project, the excitement

reflected an intriguing discovery about the

And if life

spell "job security" for

every

truly

had been discovered on Mars,

NASA employee.

But Pathfinder had never been intended to search for

life.

The

hard-

SOJOURNER

208

ware of the lander and rover was complete, and was only

a

week or

from being shipped across the continent to Kennedy Space Center tegration with the launch vehicle.

The

so

for in-

designs had been fixed for almost

two years and were now immutable.

The response til

to the possibility of

life

on Mars would have

to wait un-

future missions. Yet JPL and

est

NASA did not ignore the firestorm of attention and inter-

now focused on

Shirley,

Mars.

The Mars Exploration Program,

had already been studying the

possibility of a

led

by Donna

Mars sample return

mission that might launch as early as 2005. Now, with the President of the

United States calling for an international

mine the appropriate next Mars,

scientific

NASA headquarters wanted to know if the pushed forward. Could

ple return could be

conference to deter-

steps in response to the evidence of past

life

on

schedule for Mars sam-

a mission

be ready to launch

by 2001?

A

small group of engineers at

JPL found

itself

running around in a

mad rush to try to answer that question. Most of the rover team was blissfully ignorant

of

this activity.

MFEX

But Jake Matijevic was the

Manager, and could not stay out of the

fray.

Rover

He was the only one available

who

could imagine what a rover that could collect Martian rocks would

look

like,

and what

it

would

cost to create. Matijevic shielded the So-

—one

journer rover team so that they could make today's mission

hardware instead of paper

For the final

ule

showed

were

fifty-nine separate tasks to

to a plan

left

schedule,

The

—a success.

two weeks before Sojourner left for the Cape, the rover sched-

tests, verifications,

worked

built of

be completed. Most of these were

and calibrations of the

mapped

to the day,

flight rover.

effect

rover

team

sometimes to the hour. Weekends

open, and marked as "contingency days."

we would be working seven

The

days a

week to

If

we

got behind

catch up.

of the pressure was pervasive throughout the rover team.

We all tried to make allowances for each other's peccadilloes and lapses. We were still held together by our shared objective to put a working microrover on Mars.

Some

people handled the stress better than others.

Meteorites,

Working long hours was

Life,

like

and Job Security

an addiction. Once

enough, things just didn't seem right unless

It

done

it

for long

often found myself angry

I

people on the rover team for no good reason.

at

was time

I'd

was working. As we made

I

the last push toward shipping the flight rover,

and snapping

209

for a vacation.

* Once Sojourner was on her way to Kennedy Space Center,

week

off.

novels,

For most of that

watched

By Friday

night,

I

was relaxed enough a rover

The simulator was

would let me

predict

mands we gave

did next to nothing:

I

I

full

slept late, read

TV

home computer with off-hours.

week

took a

I

it,

that

I

felt like

command simulator

my pet project,

tinkering

on

my

been working on

I'd

a spreadsheet

program

in

that

how long the rover would take to complete the comhow much data it would send back as a conse-

and

quence.

On this particular Friday night,

I

thought

not yet implemented: the "Local-Time the rover

how

long to

sit idle.

I'd tackle a

There was the "Relative-Time WAIT,"

tween completing one action and five

minutes for

it

to

I

WAIT" We had several ways to tell

which we used whenever we wanted the rover to wait

then wait

command had

"Turn on

starting the next:

warm up." We

a period of time be-

could also

tell

this sensor,

the rover to

wait until a specific day, hour, minute, and second.

The "Local-Time WAIT" command

let

us

tell

Sojourner what to do

according to a Mars time clock. Since the Martian day was thirty-nine

minutes longer than an Earth

day, the

ways getting out of synch with each ference

other.

It

wasn't

between two time zones on Earth;

difference kept changing, day

Earth clock would have been ties that

Earth clock and Mars clock were

were naturally

would always be done

by

less

tied to the at

noon,

like the

instead,

day. Specifying

constant

when

day,

the sun

dif-

the time zone

times according to an

convenient for scheduling rover

Martian

al-

activi-

such as observations that

was

as

high overhead as

possible.

The

rover also had a built-in safety feature

pended on the Mars

local time. If solar

— auto-shutdown— that de-

power dropped below

a set level,

SOJOURNER

210

and the time of day was

later than specified (usually 6 p.m. Mars local would shut itself down, even in the middle of a command

time), the rover

sequence.

down

its

The

was

idea

from inadvertently running

to prevent the rover

precious, nonrechargeable batteries

by operating

for long peri-

ods at night.

wrote

I

tested

a short software routine to

do the "Local-Time WAIT."

with a "Wait until 10 a.m." command, but

it

Thinking through Mars clocks and Earth times was a bit expected the routine to operate perfectly the software code again ...

What I had planned as I

wasn t

I

tine just didn't have that

many

rectly.

piece of information

Mars in

local time

Mars

my

I

was

I

tricky,

so

hadn't

I

went through the

The software

was taking

sure that

I

a long time.

The more

had implemented bad

starting to get a

feeling.

software depended on to be able to

By

local time.

thing:

The

telling the rover

when any

tell

time

when midnight was on

Mars,

other Mars local time would occur.

stored time specifying Mars midnight

was using the same time values

memory,

so

if

the time

flight rover as well. rect.

I

cor-

The one

software was correct and the results were wrong, that could only

one

it

—was the Greenwich Mean Time that matched midnight

could then compute

ulator

rou-

of code in which a bug could hide.

more I was

So what was wrong?

time.

finding the problem.

lines

a short exercise

studied the software, the

first

I

wasn't working.

it

I

was wrong

that

I

If

it

my

mean

My sim-

was wrong.

had loaded into Sojourner's

in the simulator,

it

was wrong on the

checked the value. Sure enough, the time was incor-

changed the number to true Mars midnight, and the

results in

my

The time

for

spreadsheet immediately made sense. I

thought,

As

I

"What

a screwup!"

studied the error, a horrible scenario

came

to mind.

midnight that was currently stored on Sojourner was so

morning on Mars would seem I

far off that early

to the rover to be the middle of the night.

could imagine the rover waking up

at

around

8 a.m.

Mars

local time

on

landing day, noting that the solar power level was low and that the local

time was after

its

scheduled bedtime, and then shutting

—

itself off

all

be-

wouldn't

first command sequence came from the lander. Sojourner know that the sun had just come up, but would instead think the

sun was

setting. If the operations

fore the

team couldn't get enough telemetry

Meteorites,

Life,

and Job Security

from the rover to determine what was going on rover

was awake,

proclivity to

it

power

211

in the short time that the

might take days to understand Sojourner's sudden itself off,

an ironic reversal of the spurious wakeups

we had experienced during early testing of Marie Curie. Of course, I'd found the error. The scenario would never happen. I

couldn't stop

wondering what would have happened

to play with the simulator.

When would

I

if

I

But

hadn't decided

have caught the problem?

come across it? When I went back to work the next Monday, I rushed to announce my mistake to the team. I was then curiously disappointed when my mid-

Would anyone

else

have ever

He and

night error didn't even rate an entry in Allen Sirota's problem log. the rest just didn't see the midnight error as a problem:

rameter change to be made. From

Sirota's point

It

was simply

a pa-

of view, the rover had

never exhibited the symptom, so there was nothing to report. To me,

was

a time

bomb waiting to go

A month

off.

command was added

one extra

later,

prelaunch sequences sent to Sojourner

memory was changed to the and that bomb was defused forever.

rover's

lander, intended to fly

Cape Canaveral,

Two weeks

gus."

correct value for midnight

in the

on Mars,

and lander traveled separately

Florida, arriving in mid- August.

designed to drive, was flown across the

airliner.

Sojourner had

its

own custom

which the rover team commonly referred

The words "MARS ROVER" were

tion for the

one of the

through space, was trucked from Pasadena to

later the rover,

country by commercial container,

to

KSC. The parameter

at

For the first leg of the trip to Mars, the rover

The

it

trip, a

storage

to as the "sarcopha-

printed on each side. In prepara-

standard airline modular cargo container had been

delivered to JPL. Eisen and his crew loaded the sarcophagus into the cargo

container and carefully strapped

it

in place.

They transported

the cargo

container to the airport, then went out on the tarmac and supervised the

loading of the Mars-bound cargo into the ulations forced Eisen

aircraft.

From there,

airport reg-

and Ken Jewett to return to the terminal. They

walked through Security and boarded Sojourner's plane with the

rest

of

SOJOURNER

212

the passengers. After the flight landed in Orlando, Eisen and Jewett wit-

nessed the unloading. Sojourner was three thousand miles closer to Mars.

» When Tom Economou

APXS

sensor head,

Kennedy Space Center with the

arrived at

it

didn't

work. The

contained the radioactive sources that the ful data. Allen Sirota

enough,

it

didn't

was

work.

I

head was supposed to be

flight

on Sojourner, except

identical to the spare unit already installed

APXS needed

didn't

it

work at

all.

that

it

to get meaning-

Economou. "Sure

getting used to dealing with

mean,

flight

Nothing but noise

in

the spectrum." So they transferred the radioactive sources from the flight

sensor head to the spare unit.

The

spare

was remounted onto Sojourner,

and the APXS started working again. "The

head

identical sensor

wasn' t." The spare unit would be flying to Mars. For the

last

lander petal.

time, the mechanical

The RHUs were

He took off all MOVE BEFORE FLIGHT," each wheel.

would

inadvertently

make

team

tied Sojourner

down

to the

installed. Ken Jewett carefully locked

the

components with red

down

tags saying "RE-

ensuring that no lens caps or laser covers

the trip to Mars. Jewett

was glad

to be

done

with Sojourner. For him, one of the most satisfying moments of the project

had been back

at JPL, "putting

knowing that it was the

down Marie

last

time."

the rover together for the

The

final

hours of the morning. There was no one Jewett stopped to appreciate the

down. Every cable

wanted

it.

inside

wondering

else

moment.

around.

.

.

to tear

many more

When it was done,

just let the building quiet

"I

and outside was

if

.

assembly occurred in the early

tied

Nothing on the vehicle was out of

sleepless nights

time

Over the past year they had had

Curie and Sojourner, and then reassemble them

times than anyone had expected.

last

up

place.

exactly

where we

There would be no

we'd made any mistakes.

It

really

went

to-

gether well."

At the Cape, Jewett watched the Pathfinder team close up the

lander.

The next time anyone would see Sojourner, it would be in images from another planet. The lander was mated to its heatshield, backshell, and cruise stage. Later, Jewett found himself "watching it go on the rocket.

Meteorites,

Knowing

that there

smiled. 'And

I

never have to take

was held

toasted to a job well done. the launch

—

and Job Security

was nothing more

would

The prelaunch party

Life,

at JPL, in the

at the

apart again!"

Mission Manager's house. People

They talked about where they would be during blockhouse

Kennedy Space Center monitor-

at

ing the spacecraft, or out in the bleachers just watching.

This would be the

cited, relaxed, happy.

one

to gather together in

I

launch preparations

stood in the corner, next to the food, and spoke with Miguel San

He worked

finder, the

software that

direction,

which was

the

ACS

made

flight

tall,

blond, balding, expressive Ar-

—Attitude

Control System

sure the spacecraft

critically

was comparing Pathfinder

working on

when

important

to other projects.

software for Cassini.

We

—

for Path-

was pointing in the

firing thrusters. "It's

amazing,

were going

right

San Mar-

really.

I

to Saturn.

was

And

managed to make it boring! Everything was compartmentalized. You

they

could only

your

skills

work on your one anywhere

else. If

ested in letting you deal with

area.

There was no opportunity to apply

you discovered it

yourself."

a

problem, no one was

inter-

On Pathfinder, you just did what

to be done.

Although mostly working issues, the rover

lieved in I

final

coasts.

gentinean.

had

ex-

place, before they scattered either for a well-

Martin and his wife. San Martin was a

tin

They seemed

chance for most of the team

last

earned Thanksgiving holiday or to participate in

on both

He

could do." Jewett paused.

I

it

213

separately,

and lander teams had shared

both teams.

I

was confident

at

odds over

specific

a dedication to success.

in the ability

had worked with. But when you pause

long, questions

and often

of every team

for breath after

you cannot yet answer have time

Had any one engineer missed something vital? Had we really done everything that needed

to form.

doing?

I

be-

member

running for so

Would

it

work?

PART

3

GOING TO MARS

NINETEEN

EVEN

A

JOURNEY OF

THOUSAND MILES

A

.

day before Thanksgiving, four days before the appointed hour,

The already knew we had an 80 percent chance of scrubbing the

we

first

launch attempt. The rover was ready The spacecraft was ready So

was the launch front

moving in

at

But the long-range weather forecast showed

a

Cape Canaveral. Rain and high winds would erode our

The project could not risk the rocket being destabilized like

safety margin. a

vehicle.

motor home being battered around on the highway

in gusty winds.

One unchangeable rule of solar system exploration is that the planets don t wait. They move inexorably in their orbits and cannot be cajoled into match those of mere humans. Only through

adjusting their schedules to cleverness can

we

leverage the limited

lift:

capacity of our rockets and the

occasional alignments of the planets into opportunities to send spacecraft across interplanetary space to actually find a planet at the other end.

However, with some extra propellant leeway.

We

December months

had

we would

Mars and Earth were

have to wait another twenty-six

in position to try again. Realistically,

not going to pay to keep the entire operations team active for

two years doing nothing. So not launch

you can buy some

chance to launch once each day from December 2 to

31. After that,

until

NASA was

a

in the tanks,

at

all.

if

we

did not launch in December,

we might

SOJOURNER

218

—the periods of time get the spacecraft where needs to go— are

For interplanetary missions, launch windows

each day that a launch will critical

and

it

During the

fleeting.

first

two-minute launch window each idly shrink

down

to only

pacity of the rocket

two weeks of December, there was

day. After that, the

one second each

would just not be

day.

window would

By January,

a

rap-

the energy ca-

sufficient to get to Mars.

When Pathfinder was launched, it would first go into orbit around the Earth, and then be given a final boost to put

it

on

its

way

to Mars. If only

the Earth's equator were exactly aligned with the plane of the solar sys-

tem, then Pathfinder could be launched at any time of day, and boosted into

its

Mars trajectory whenever

was headed

it

in the right direction.

reached the point in

But the Earth's

twenty-three degrees from vertical. So to

make

full

is

when

plane in which stant,

all

the Earth's motion

the planets traveled.

when

tilted

we needed

would hurl Pathfinder along

And

this

it

about

use of the speed ad-

vantage of the thousand-mile-per-hour spin of the Earth,

launch only

orbit

its

axis

happened only

for

an

to

the in-

twice per day.

For half the day, the Earth's spin would hurl Pathfinder out above the plane. For the next half of the day,

it

would hurl Pathfinder out below.

Only during the brief period of crossover would the spacecraft be hurled perfectly along the plane,

on

target for Mars.

Those short moments were

the launch windows.

There was yet more complexity space probe

The

sent to Mars,

is

it is

current position of Mars

craft gets there,

Mars

is

will surely

in the planning of a launch.

Mars

the next problem

is

a crater

The

not the target, for by the time the space-

be gone. Instead, the spacecraft must

ity

will arrive there. If

at a specific point

you can master

slowing the probe upon arrival so that

and

set

at the

that objective,

it

doesn't

make

planet and spacecraft meet.

window was a two-minute period centered on 2:09: 1 1 a.m. December 2. By the morning of Sunday, December 1, the probabil-

first

EST,

when

a

not a simple case of "point and shoot."

out on a path that will cross the orbit of Mars precise time that

When

launch

of being forced to scrub due to weather was up to 90 percent, so the

decision

ber

3.

was made

Even a Journey of a Thousand Miles ...

219

to hold off until the next launch

window, on Decem-

The weather

forecast said that,

much improved, with an Since Sojourner for

sleep through the launch, there

our team to do during the

the rover system

were

all

team had

wanted

in town. For

to

final preparations.

be

was

most of

to see Sojourner

would be

third, conditions

80 percent chance of a "go" for launch.

would

the lander /rover system

by the

Only

was nothing

few members of

a

in Florida in case a serious

problem with

we

identified before launch. Nevertheless us, this

was

to be our

launch. All of us

first

on its way.

With the launch postponed one

day, the

postlaunch reception on the

patio of the Patrick Air Force Base officers'

became another

club

prelaunch party. The base was only a few miles south of Cocoa Beach, so

you could look north over the water

to

where the Delta rocket shone

bright white in the crossed floodlights that illuminated ther side of pad

launched.

C17A

to be seen, poised to begin

its

and yet the reason

for the party

was there

journey.

for Mars. After a bit

I

stood there and talked with Jack Morrison, looking

we thought we had found it,

bank. Just a star with a red tinge to miles away. Tonight

The

ei-

rolled onto the beach, barely visible in the light spilling

over from the patio.

I

Beacons on

blinked red, signaling a payload ready to be

We were miles away,

The waves

it.

we were

it,

rising

above a low cloud

a point of light a

hundred million

going to send something there!

looked from Mars to the rocket. Back to Mars. Back to the rocket. origin of the trip

was

in view.

behind me. The starting gun was

So was the destination. Music played

set to

be

fired in just seven hours.

The launch was scheduled for 2:03 a.m.

There were three viewing way, and Jetty Park.

sites:

the

VIP

bleachers, the

NASA

The bleachers and causeway were within the

Cause-

confines

of the Cape Canaveral Air Station, north and west of the pad. You needed a car pass to get

onto the base; they wouldn't even

hour before launch, and people would spots they

let

you

in until

one

be jockeying for position in the

were allowed.

Since the launchpad

was near the southeast corner of the

base, Jetty

SOJOURNER

220

—outside the base, where anybody could go—was actually the

Park

est site to the pad.

The morning before

the scheduled launch,

clos-

we had

scouted the area. The jetty had been built of rocks piled on each other.

It

reached a few hundred yards east out into the water, topped with a nar-

row concrete walkway widened jetty

to

railed

make room

on both

for sinks

sides. Periodically the

walkway

—so that the fishermen who lined the catch—then narrowed again. In day-

day and night could clean their

light,

we

could see pelicans sitting watch from every available

piling.

When they took flight, it was to gather around a pair of porpoises making their way out to sea. A colony of feral cats seemed to be living among the rocks of the jetty

itself.

As we approached, they grudgingly gave ground,

seeing us as interlopers on their

turf.

The fishermen ignored us.

From the jetty, only the bottom third of the rocket was obscured. And because the jetty was so long and narrow,

it

was

unlikely that anyone

would block our view.

The

rover

night, the

the

The

jetty.

team had agreed that Jetty Park was the place

convoy of rover team members, line to get in

was

short,

but there was a

charging the usual dollar per car to park. there

was someone there

at

I

to be.

At mid-

and friends headed

family,

line.

for

They were

wondered distractedly whether

midnight on non-launch nights to take your

money There might have been a hundred people out on the jetty before but there was plenty of space.

We

established our personal spots

us,

among

the others. Jack Morrison and his wife had brought a pair of binoculars.

We

passed them around, taking turns studying the Delta rocket.

beautiful.

To the naked

Through the

On it were

ish blue. las,

eye, the Delta

binoculars,

you could

had just

see that

it

seemed

was

a

It

was

tower of white.

actually mostly a gray-

the symbols of Pathfinder, and of

McDonnell Doug-

manufacturer of the Delta.

We

waited.

It

Nervous energy,

We

waited.

Russians'

Mars

I

was cold and damp. Yet some people were

in T-shirts.

supposed.

We

talked.

I

paced.

'96 mission that

told stories, thought about the

had ended

at the

bottom of the ocean

we were nervous. With excruciating slowwindow was grinding toward us.

only two weeks before. Mostly, ness the two-minute launch

We

Even

a

Journey of a Thousand Miles

.

.

221

.

A groan spread through the group. "They scrubbed the launch! A conkept staring at the sole! A damn console!" The crowd began to disperse. I

rocket.

didn't

I

other moment,

want I

to believe

wasn't going to happen. But after an-

it

knew the launch window must have closed. A minute be-

fore the launch, a software glitch

had occurred

ground computer

in a

monitoring telemetry from the Delta. Not enough time to

window

fix it

before the

expired.

Tonight would not be the night.

We were back the next night. But there were fewer of us. The fishermen were grumpy about all the people scaring the fish. For tonight, the launch window would be slightly earlier, at 1:58 a.m. The Principal Investigator for the APXS came by. "So, Andy. What's the reason going to be tonight?

him

there

was

right.

would be no

About

someone

a

Why won t we

reasons, that tonight

a voice shouted.

could only watch.

countdown.

We

I

all

launch.

stared at the Delta.

And

the rocket

I

quiet.

Then

I

told

hoped

I

I

heard

the base

was moving. "That's

Everyone cheered and applauded. Almost everyone.

I

willed the rocket into the sky.

There was no sound yet from the climbed into the

sky,

didn't. All the cliches

so bright that

rising rocket. I

thought

I

It

was too

my vision where my

far away. It

should look away, but

about "bright as the sun" were true.

gion around the flame in small part of

we would

this time?"

minute before the launch time, people got

reciting a

of the rocket got suddenly brighter. it!"

launch

retinas

I

could see a

were

re-

saturated.

A

my mind wondered if I'd see all right after this was over.

The crowd spontaneously cheered again. I wasn't sure why, except maybe that it was clear that the rocket was flying true. The sound reached us finally across the water.

A kind of staccato roar,

voice of pure power.

I

of ex-

couldn't see the Delta itself anymore, just the flame and the

trail

haust that drew an arc between

was aimed

straight at the

The flame was tered briefly,

it

and the pad.

It

looked

like

it

moon. smaller now, but

it

seemed just

dimming and brightening

again.

as bright.

Then

it

stut-

There was the shortest

SOJOURNER

222

pause, and the six ground-lit solid rocket boosters

peeled away from the Delta, their ers

had taken

over.

The flame of

became

work complete. The

visible as they

three

air-lit

boost-

Another cheer. rapidly

becoming

jettisoned solid rocket boosters

had been

the engines

was

a bright red star in

the sky.

The formed

a flickering constellation.

I

knew

that they

left

must be on

moment they seemed a new Pleiades.

to falling into the sea, but for the

the sky, six twinkling lights,

Only one minute as well.

later,

the

air-lit

solid rocket

two minutes

though some others

into

its

in the

to

They

their

way

be hanging

in

motors were jettisoned

me to discern the boost-

But the Delta was already too distant for

ers falling away,

behind.

crowd claimed they could. Just

mission, the spacecraft

was

traveling over a mile

and

a half per second.

The

spot that

lost sight

of

was Pathfinder was

hundreds of miles away.

down

bottles

their faces

were coming and

Some

out.

didn't even realize

it.

people had tears

The

rover crew

gathered as most of the other watchers headed back toward their all

I

it.

The champagne streaming

tiny now,

shook hands. "Good work! Good job!" everyone

take credit for

what I had just seen? I wasn't

Allen Sirota raised a cup. "To the rover!"

said.

cars.

Could

I

We

really

sure.

We all drank the toast.

We started packing up to follow the rest back along the jetty. Again we all

shook hands.

I

found myself repeating the same statement to

everyone: "We're in business!"

With Pathfinder on plete.

Mine was just

its

way

starting.

to Mars,

many

people's jobs were

com-

TWENTY

CRU1S1N'

Pathfinder was cruise

on course for a Fourth of July landing on Mars.

was spinning

stage

we

planned. But by the time

was already ical to

in trouble:

its

crit-

there to measure the rotation rate of the space-

approximate orientation in space. The sun sensor had

sensor heads situated on the rections. Sensor

heads

heads 4 and

5

1, 2,

more

puter with

precisely, its

body of the

and

five

spacecraft, facing in different di-

3 faced to the sides

looked back along the spin

tracker onboard, w^hich tion

the onboard sun sensor,

navigating the spacecraft, was half-blind.

and

stage;

twelve rotations per minute, as

got back from Florida, the spacecraft

When first activated,

The sun sensor was craft,

at

The

would determine the

of the spinning cruise

axis.

There was

also a star

rotation rate and orienta-

but only once the sun sensor had provided the com-

rough estimate of

this

information.

To make course corrections during the

cruise to Mars, the spacecraft

computer would require precise knowledge of the direction the spacecraft

was pointed.

things worse, and

Firing thrusters in the

wrong direction would just make

would cause Pathfinder

to miss

The software team manager explained

Mars completely.

the impact of the sun sensor

situation matter-of-factly: "We're not running the Attitude Control Sys-

tem now, and we

can't until

we

can use the sun sensor. We're almost

cer-

SOJOURNER

224

tain that there's debris obscuring the optics of sensor heads four If

we

could send somebody

could

fix it in a

minute. Not an option.

end of the mission. lem.

We

wise,

we

think

can

In the

we

all

Do

I

some

out there with a tissue and

think we can

have a solution.

If

we

can

five.

alcohol,

the problem,

it's

we the

We're working the prob-

fix it? Yes.

If it

t fix

and

works, everything

is

okay. Other-

go home."

meantime, since the Attitude Control System, or ACS, could

not be turned on without the sun sensor to guide

it,

team members

in

other subsystems identified alternate means to determine the spacecraft spin rate and orientation.

By examining the change

radio signal transmitted from Pathfinder as cations engineers

came up with

rotated, the

estimates for both.

team derived estimates of the

tion

it

in the strength of the

telecommuni-

The planetary

spacecraft's distance

naviga-

and speed by

sending a radio signal to the spacecraft and waiting for the precisely timed response.

The telecommunications and navigation their results

each morning

at the status

representatives presented

meeting.

Though independently

generated from separate data sources, the estimates of both teams were a nearly perfect match.

The story began circulating through the ground team that everyone knew more about the orientation of the spacecraft than Miguel San Martin,

who was

the engineer responsible for the ACS. "Even the janitor

knows more about

went the joke. Of

the spacecraft attitude than Miguel"

course, San Martin didn't yet have a

working sun sensor to

rely on.

It

may

have rankled some that two other subsystems were reporting the infor-

mation that should normally be coming from him, but San Martin took the situation in joke.

The

good humor.

In fact, he

was the

first

person to

tell

data from the telecommunications and navigation teams did

low the project

to track the state of the spacecraft, but

attitude control, since the estimates

were

based on ground processing of large data

would be so

far

away

it

was

sets.

Eventually the spacecraft

from the ground would

take over ten minutes to arrive onboard. For course corrections and

in real time.

And

its

al-

useless for

after-the-fact determinations

that any instructions sent

neuvering, the spacecraft needed to assess

board

the

ma-

own orientation and spin on-

that required the sun sensor.

225

CruisirY

Since the internal electronics of the sun sensor mally, the theory

was

that debris

was functioning nor-

was blocking the sensor heads. The most

source of the debris was contamination from the protective shroud

likely

that covered the spacecraft during launch.

Once

in space, the

the shroud had explosively separated, as intended, falling

two

away

parts of

to either

Perhaps during separation some propellant had spat-

side of the Delta.

tered onto the sun sensor heads.

on only

Attitude control for Pathfinder could not be shifted to depend the remaining functional sensor heads. in

in a direction different

from the

others.

would track the sun with the

of the

trip to

Working

Sensor head

5

was generating no

since

it

seemed remote.

signal.

at least that

would have

weaker than

it

a signal that

was not

likely to

was supposed

to be.

from head

Even such

a

could do the job.

and the software rejected the data

but with head

and the

5

4 could

4 confirmed that

weak

is

it

was four times

signal should have

But the onboard software was refusing to

tude of a sensor head's output dropped too low,

made

cleaning

improve. So the

use the head 4 data. This was a designed-in safety measure:

heads. This

most

solar panel in

what head

looked normal, except that

to track the sun.

sun sen-

to be the assumption,

to function with

provide. Careful review of the data

was putting out

had

5

its

The chance of it spontaneously

Similarly head 4

Attitude Control System

on,

heads

spacecraft in any orientation. For

Only sensor head 4 or

was gone. Or

itself

been enough

in concert, the

Mars, the spacecraft had to be oriented with

the general direction of the sun.

now

five

they were not truly backups to each other. Each head was pointed

all,

sors

Even though there were

it

If

the magni-

was considered suspect,

in favor of that available

from the other

sense in the case of a problem with a single sensor head,

out of the running, there was no other data to

fall

back

safety protection feature prevented the Attitude Control Sys-

tem from operating on

the only

good data

The lander software team wrote

it

had.

a modification to the

onboard

soft-

ware, a "patch" that would cause the Attitude Control System to accept

weak signal from sun sensor head 4. This patch would replace a small piece of the program running on the spacecraft, to change the conditions under which sensor data would be rejected. The new software was first tried out in the spacecraft testbed. With the weak signal simulated, and all the

SOJOURNER

226

other aspects of the spacecraft configuration duplicated with delity the

ground team could muster, the patch functioned

all

the

fi-

as expected in

the testbed.

On December 7, four days after launch, and

installed

came

into

on the

JPL

actual spacecraft.

to handle the job.

It

No

the patch

was

was

a Saturday,

to be uploaded

and a small group

one expected any problem: They'd

just uplink the data files containing the patch, install the patch,

and

start

running the Attitude Control System. The whole process should take two hours.

But when the Pathfinder team uplinked the

files,

they didn't go

through. After repeated attempts, the spacecraft continued to reject the

What was going on? They determined that very short data files and individual commands were being received onboard, but nothing else. files.

More and more members of the word of the problem got out. They tried slicing the original

Pathfinder

data

files

sending those. This was partially successful: board, but others did not.

They

team were showing up

as

into many smaller files, and Some of the files arrived on-

didn't understand the cause of the

still

And it was getting perilously worse: Sometimes even tiny indicommands weren't reaching the spacecraft! If they couldn't com-

problem. vidual

mand the

spacecraft, the mission

The answer

finally

was

came from

over.

the telecommunications engineers.

They had been monitoring an unexplained communication five

minutes.

signal

from the

dip in the strength of the

spacecraft, a short dip that recurred every

Now what would cause that? The spacecraft was rotating at

twelve revolutions per minute, or once every five seconds.

The mea-

surements they had been taking of signal strength weren't continuous; the telecommunications engineers realized that the dip was actually happening every five seconds, but they were only catching it in their data every five

minutes.

Some

piece of metal

time the spacecraft rotated.

They were To

getting

get data

up

dealing with very

chopped

was blocking the radio

No wonder the

files

signal

once every

weren't getting through:

off in the middle of transmission.

to the spacecraft, they applied the usual technique for

weak

radio signals, one that should never have been

22?

Cruisin'

needed with Pathfinder

The

rate.

normal

data

files

still

so close to the Earth:

were uplinked

at 7.8 bits

They dropped

the data

per second, one-thirtieth the

rate for this point in the mission. This time they got through.

The

patches were installed. After a harrowing twenty-hour delay, the vate the Attitude Control System, and

it

more toward

sent to acti-

was up and running

time in the mission. The Pathfinder team to turn a bit

command was commanded

for the first

the spacecraft

the Earth, and the communications problem

went away.

A few days later the spacecraft was "spun-down" from 12 rpm to 2 rpm. At

this rotation rate the star tracker

would

operate.

With

the spin rate

computed from the working sun sensor output, the ACS was "Celestial" stars.

mode. The

Pathfinder

star tracker

shifted into

began scanning, and locked onto two

now knew exactly where

it

was

pointing.

Miguel San Martin was once again the principal source of spacecraft orientation status reports.

Sojourner had ridden through the sun sensor campaign in oblivious sleep.

The

first in-flight

healthcheck, through which

rover had survived the launch unscathed,

about two weeks

we would

was planned

for

verify that the

December

17,

after launch.

This was going to be our

first

moment of truth.

There was no reason to expect that there would be anything wrong. Ever since the sun sensor that the only times to

fix,

I

had been

relaxed.

I

had convinced myself

be worried were launch and landing.

be working. Until Allen Sirota came

It

seemed

to

by a few days before the healthcheck

and talked about being nervous. Other engineers dropped by with

their

own ideas about which rover components might fail. So about twenty-four hours before the rover wakeup, my stomach started churning. What if we ordered the lander to power on the rover, and all we heard was silence? What would we do then? How could we fix it without any data? Every member of the rover team had shown up for this first postlaunch check of Sojourner. They each knew how their particular subsys-

SOJOURNER

228

how they were built.

terns functioned,

to interpret the full telemetry stream. as

it's

This

supposed

to,

we

not a problem.

is

I

of them

all

warned them:

knew yet how

everything goes

"If

should receive exactly one 'Error Report' message. It's

not really an error.

check the accels and note that

will

But not

When the rover wakes up, it

not seeing any gravity That will

it's

an automatic mission phase change from 'prelaunch' to

trigger

phase. Every time the rover does a mission phase change

sends a telemetry report to

tell

us what

it

did,

on

it

and that report shows up

as

an error message." Someone asked which type of error we would error types were

we saw

'cruise'

own,

its

see. All

numbered in hexadecimal code. I looked it up. "1D02." second one, then

a different error message, or a

we would

If

have

problems.

For the telemetry that would be generated by the healthcheck com-

mand Allen

Sirota

and created

a checklist of the numerical values expected for each of the

and Art Thompson had polled the various subsystems

how

far the

channel could deviate from

the expected value without raising an alarm.

The expected numbers were

hundreds of data channels, and

based on the values observed prior to launch, together with the few changes expected due to the postlaunch environment. The primary changes

we

anticipated

were

in Sojourner's

onboard temperatures, and in

the rover's gravity readings, which should be virtually zero, since the rover

was

Just itself.

in free

it

The change

Mars.

to "cruise" phase

landing on Mars. All

be doubly

safe,

command sequence

Once

the rover

was

was the most

we needed

the radio

the right

we

to

ever, if the rover

do

between

for the rover to

it

know

up sometime during

also told the rover to switch to "cruise"

for the healthcheck.

in "cruise,"

it

would be ready

it

to switch over to

next

was never powered on during

Sojourner would wake up on Mars

with no

to

wake

critical step

woke up and sensed gravity again. So even if modem was damaged in the landing, Sojourner would still do thing and perform its preplanned contingency mission. How-

"on-Mars" whenever

failed,

way to

was no longer on the Earth was

cruise. Just to

in the

its

powering on the rover was more important than the healthcheck

now and the that

on

fall

way to determine

that

it

was not

still

still

on

cruise,

in

its

and the

modem

"prelaunch" phase,

Earth. In the

wrong mis-

229

Cruisin"

sion phase and with

no communication,

eternity, or at least until

back

some

future

it

would

sit

on the lander

Mars astronaut picked

it

up

for

all

to take

it

to the Smithsonian.

Thompson was

Art project

team

the rover engineer assigned as liaison with the

for the day's healthcheck.

I

sat

with him in the Mission Sup-

port Area, the Pathfinder equivalent of Mission Control. The

Sun computer workstations, and

tained a dozen

Communications Assembly tion to

VOCA

room

Voice Operated

units that enabled the operators at each sta-

communicate with each other over the voice network. The

the rover

team was

con-

down

situated

the hall at their

own Sun

rest

of

workstations

with appropriate telemetry displays. They were also hooked into the voicenet. but

it

was Thompson's job

The operations team would

Pathfinder ground operations team.

Thompson on

to represent the rover to the refer to

the net simply as "Rover.'"

As the team worked through each step of the procedure to prepare the spacecraft to support the rover healthcheck. the Flight Controller

polled each position to ensure that

we were unanimously

ready to con-

tinue.

Back

in

November, we had uncovered

that caused the craft

computer

to reset

a

bug

in the lander's software

—temporarily bringing down the space-

—whenever the Attitude Control Svstem was running and the lander

and rover were communicating.

Now we

reached the step in the proce-

dure where the

ACS would be

shut

San Martin was

at the Attitude

Control System position.

The tions.

flight

The

down

to avoid this problem.

Flight Controller

space. Telemetry

go."

approved the uplink, and off went the com-

came

back,

command, then

first

mode. ACS

is

offline.

2 million miles of

confirming that the spacecraft had

re-

command had been executed. "ACS confirms we are now in super-

indicating the

San Martin reported through the net. idle

is

engineer also confirmed he was ready to send the instruc-

mands, through the Deep Space Network and across

ceived the

ACS

Miguel

No problems."

Richard Cook, the Mission Manager, had been standing

watching the goings-on. "Okav. Miguel. Everything shut down."' Off the net. he continued mischievously. don't need vou. You're fired.

Go home."

in

back

ACS is "Without ACS we

is

fine,

and

SOJOURNER

230

San Martin stayed.

If

we

followed the procedure,

ACS would be

run-

ning again in fifteen minutes. Slowly, carefully,

were ready

we worked our way through the procedure until we command sequence that would wake the

run the lander

to

rover and send

it

commands. "Rover is go

The command went

up.

"Don t

for

sequence activation."

commented

reset us, rover!"

Flight Controller, to the ire of the rover team.

It

went across the

the

voicenet,

probably unintentionally.

The lander sequence turned on

modem, We knew the se-

the lander-mounted rover

the lander's side of the communications link to the rover.

quence would wait ninety seconds and then cause the lander to apply

power for ten seconds to rover.

using to

The its

relay

a

magnet located just under the

activate,

its

reed-relay

on the

and the rover would begin to power-up

own onboard battery. The rover would take

go through

light, a

would

startup process, then send

its initial

about thirty seconds

At the speed of

data.

message would require about thirteen seconds to cover the

tance to the Earth

.

.

dis-

.

There was the error message. Sojourner

lives!

"We've got rover

now in mission phase 1, cruise phase. There's the healthcheck. We're preparing for APXS operation. Rover's now in a comm hold for the APXS. The next command we see should be number 2010, then 2009. There it is. Out of comm hold. telemetry!"

It

was the

right error message: 1D02. "Rover

is

Now we're reading out memory. Yes. Yes. Rover has shut down!" We had just had a conversation with our rover, which was almost ten times farther away than the

Moon, and

it

had worked just

like a

ground

test.

While Miguel San Martin oversaw the reactivation of the spacecraft ACS, the rover team met to review the telemetry

and current readings matched. ported status as "Good."

thought we would lander's thermal tions

see.

matched.

All eighty devices

we wanted

onboard the rover

Some temperatures were lower

re-

than we'd

But our expectations turned out to be based on the

environment from yesterday, not

were wrong, and the

crepancy

in detail. All the voltage

rover's telemetry

was

today.

So our expecta-

right. Just the

to have. All the voltage

kind of

dis-

and current readings

231

Cruisin'

With the healthcheck complete,

I

could write the

port for the Mars Pathfinder telephone information

This morning the operations team first

woke up

first

rover status re-

line:

the Sojourner rover for the

time since launch. The rover performed an internal health status

check, accepted

command

sequences provided by the lander, operated

the onboard Alpha Proton X-ray Spectrometer, sent telemetry, and shut itself

down

as

that the rover

planned to conserve is

healthy,

rover will remain

with

powered

all

its

batteries.

We

are

happy

to report

subsystems functioning normally. The

off until the next rover healthcheck, not long

before landing on Mars.

When

I

whiteboard.

on

its

face

got back to

my

office the next day,

Someone had drawn

and

its

APXS

"tail"

I

found a message on the

a caricature of Sojourner,

held straight up in the

were four words above the drawing:

air,

with a smile

wagging. There

"WE HAVE A MISSION!"

TWENTY-ONE

WHAT ARE YOU GOING TO DO FOR THE NEXT SIX MONTHS?

SO

having fun!"

They're

concentrated, focused on

wave.

I

film.

missed

I

The dark shapes were

it

the

first

them in an attempt to capture it on

time, snapping the shutter a second too late.

But more were swimming farther out. in,

but they weren't

ideal,

right.

No

I

was

patient.

Some waves

rolled

novices these. Finally, the conditions were

new swimmers

and two

definitely catching that

rode the wave

perfectly. Click.

Sea lions

surfing at sunset.

Eight of us sat on the rocky beach of North

Seymour

Island feeling

mammals appeared to be. I was in the from home and work, too isolated to hear

nearly as carefree as the marine

Galapagos,

six

thousand miles

any news or solve any rover

and no pager.

I

crisis

should

it

arise.

There was no telephone

was obeying the maxim "Never worry about

things

you

cannot change."

Each morning our boat the panga into shore. feet to avoid stepping

years,

From

new island landing site and we rode

we

hiked around, having to watch our

there

on animals

had forgotten to

performed

sailed to a

their courting

fear

through isolation for millions of

man. Seabirds

dance ten

basked in the sun, looking more the hot equatorial

that,

like

feet

—blue-footed

away Land and

boobies

sea iguanas

rock formations than living beings. In

midday hours we snorkeled, keeping cool and observ-

So What Are You Going to Do

more

ing even

afternoon,

Next Six Months?

233

creatures apparently oblivious to our presence. In the late

we went

night, the boat's

we

with flavors

for the

new

ashore at a

cook produced from the

we

loved but sensed

all

and explored

locale

further.

Each

tiny galley prodigious meals

could never duplicate.

A life of leisure was what most of my friends and relatives imagined could

the six

live for

months before Mars: vacationing,

plain lazing about. Pathfinder

traveling,

I

and just

had been launched; the hardware was on its

What could I possibly need to do until it landed? Friends way to Mars knew how hard had worked, and for how long. Certainly there was .

.

.

I

plenty of time for a well-deserved break.

The

reality

would be ten

trip to the equator,

had built the

I

Then,

days.

would return to JPL

rover, tested

its

after this

to

make

software, launched

learn to understand the nature of the beast.

teenager

who

were we good

On the rover tions.

has just passed his driver's

we were ready. We Now we would have to

sure

it.

The

test.

one sanity-restoring

rover

We

team was

had our

like a

license,

but

drivers?

side,

I

was the one focusing on how we would handle opera-

During the integration and

had begun to sound

like a

test

phase of the previous two years,

broken record with

my repeated warnings

I

re-

I

had been

forced to defer to the hardware issues requiring resolution. But

now was

garding the need for operational and performance testing, and

my time. Jake Matijevic had made me the rover mission operations engineer, and my job was to build the operations team. Most of the system team, plus the subsystem Cognizant Engineers, were about to become the core of this the

new

team. But

many

weeks following launch primarily burnout

close to

we were

in a

new

well.

I

close

Mars was, or how much we

relax.

Now,

a

of a job well done to an attitude of a

afraid that the

months before

tions team.

time to

We had all come

month

after launch,

phase of the project. The rover team had to move

a sense

1997, six

as a

in the previous four years.

from

was

of those same people had pictured

team did not yet have

landing,

all

we

had yet to

new job

to be

done

a feel for just

how

learn.

So on January

7,

kicked off the training of the opera-

SOJOURNER

234

What were operations going to look like? Once Sojourner was on Mars, the rover

team would have three things

to do, day-in

and day-out: analyze the

data coming back telling us what the rover had done, write the command

sequences telling

The

rover

it

what

team would

The job of

to

do

next,

and coordinate with the lander team.

divide in three to get these jobs done.

assessing the health of Sojourner

would go

members from each

neering Analysis team, composed of

to the Engi-

rover subsys-

tem. This downlink team would be led by the Data Controller,

had

responsibility for

the rest of the

who

also

massaging the rover telemetry stream into a form

team could

evaluate,

and

for

documenting the

results

of

team

to

the team's analysis.

The Rover Coordinator would represent the

rover operations

the project. In the Mission Support Area (MSA), each key lander subsys-

tem was

allocated

one computer workstation, each marked by

a sign hang-

ing from the ceiling above: Power, Propulsion, Navigation, Flight Director.

Over another workstation would be ordinator

would relay requests from the

and report key rover

The

a sign

status information

rover's uplink

marked "Rover." The Rover Co-

Flight Director to the rover team,

back to those

in the

MSA.

team would be two engineers charged with

forming the desires of the science team into commands to the

trans-

rover.

The

Rover Driver would peer intently into the stereo display of the Rover Control

IMP images to see where it was safe for the specific commands to get the rover to a se-

Workstation, assessing the

rover to go, and generate the lected target. In the

of the

command

meantime, the Sequence Planner would build the

sequence, including imaging, experiments, and engi-

neering "housekeeping" functions, merging in the traverse

when

ready.

rest

The uplink team would submit

commands

the final sequence to the lan-

der team for transmission, and document what they had done.

The schedule

for

all

of

this

would not follow

a nine-to-five workday,

or any other kind of Earth day. Instead, mission operations would be driven

—called a

by the Martian day

"sol"

—twenty-four hours and thirty-nine

minutes long. Since Sojourner's day was tied to the sun, so

was

ours,

Earth-bound though

we humans

rise

of the Martian

were.

The operations

So What Are You Going to Do

team would for

live

on "Mars time": Every

work forty minutes

later

sol,

the lander

Next Six Months?

235

team members would

day,

would look something

would downlink

like this:

and science

the rover's activities earlier that

Meanwhile, the lander's

would provide

at the

end of

stereo

its sol's

worth of

of where

traverse.

would

inspect those images. This to

sol.

IMP images

after-

The

results,

generated during

own telemewe expected the rover to be

rover downlink guys

them how

tell

Late

to Earth the rover's telemetry

that included both engineering data

try

arrive

than the day before.

After landing, the process

noon each

for the

close the rover

would

had come

intended target. The downlink team would compare the received

its

rover data with the plan as

rover executed sages?

Was

forming

The

all

of

its

embodied

commands

in the

command sequence. Had the

properly?

Were

there any error mes-

Were

the rover radios per-

team would have about three hours

to interpret the day's

the rover getting too hot?

Too

cold?

as expected?

analysis

w as to

worth of telemetry and diagnose the health of the

rover.

The

whatever the uplink team would need to

know

before building

identify

command

the next sol's

The sis

idea

T

sequence.

entire rover operations

area for a crossover meeting.

team would crowd together

in the analy-

The downlink team would tell

the uplink

engineers and the Coordinator what they had discovered. For the uplink

team shift

link

it

was "morning" no matter the

actual time:

They had just come on

and needed to be brought up to speed. After the meeting, the down-

team would go

their analysis.

off to put together a

Anyone on

the project

web page

reporting the results of

would then have immediate

access

to the information.

Scheduled soon

after the crossover

meeting would be the Experiment

Operations Working Group meeting. Here, the Pathfinder ported their

latest findings

and planned the next

the science instruments, including the rover.

around to the

rover, the

of the rover's

status.

scientists re-

sol's activities for all

When

Data Controller would present

of

the meeting got a brief

summary

Jake Matijevic, as Rover Manager, would describe

a

preliminary plan for the rover's operations. Working Group members would propose changes to the plan, adding experiments and picking new

SOJOURNER

236

rocks as targets for the rover. At least one

team would always be present nario

we were

of the rover uplink

meeting to ensure that the rover

at the

mean saying no to the

meet-

scientists at the

what they proposed was too ambitious or too dangerous. One of

the engineers' responsibilities

ence another

was

to ensure that the rover lived to

With the rover scenario

in place, the uplink

team would go as a

list

of

The sequence would tell the

tian sol; building

it

command

would take hours. During the Working Group meeting,

the scientists

the next

those ten activities

for the rover.

IMP camera

the

rover what to do for an entire Mar-

might have come up with ten or so

sol;

sci-

to work.

activities,

Rover Driver and Sequence Planner started construction of the sequence.

do

day.

While Matijevic formally wrote up the scenario

mands

sce-

about to be charged with implementing could actually be

done. This would sometimes ing, if

member

would

activities for the

The uplink team had to

figure out

where

to get the "end-of-sol" snapshots of the rover.

would be used to update the

rover

on

hundreds of com-

translate into

to point the

The images

rover's position as the starting point for plan-

ning the following sol's traverse.

We needed to get the camera pointing inIMP camera soon picture-taking into their own

formation to the lander team members sequencing the

enough

to give

sequences.

them time

to build the

When the rover commands were complete, it was the Sequence

Planner's turn to write a

web page

to

document the sequence

downlink guys would know what to expect when the next

sol's

so the

telemetry

came back from Mars.

The

rover uplink

rover ever

woke up

team would

to the

new

finish its

sol.

The

job and go

home

drivers of the rover

before the

would

sleep

while the rover roamed, separated by both time and space.

As the Earth rose and rover would

Network

in the Martian sky,

commands

flash across interplanetary space

destined for lander

from the Deep Space

station at Goldstone, Canberra, or Madrid. In the early

ing, the rover

would awaken, and

find the lander waiting with

morn-

new

in-

structions for the sol.

For most deep space missions, engineers built over days or weeks.

On

Pathfinder,

we would

about seventeen hours. The whole

set

command

have to do

it

sequences

every day, in

of events had to stick to the

So What Are You Going to Do forthe Next Six Months?

timetable. tle

A delay early in the process

23?

could leave the uplink team too

lit-

time to have a sequence ready for the morning transmission, or worse,

might introduce errors that could put the rover

at risk.

Getting behind

schedule would result in lost sols on Mars, idle time during which no useful science

would be done. No one wanted

days on the surface, because there was no

be our

way

any of the precious

know which

to

sol

would

last.

So when landing day came, days a

to waste

week for

as

long

as

it

we would be

operating the rover seven

and the lander survived. Losing days of

activ-

we could get a weekend off just was not going to be acceptable. Yet if we worked seven days a week without a break, we would begin to make mistakes, let alone hate our jobs and become ity

on the Martian

surface so

strangers to our families. .And those mistakes might well hasten that eventual failure that

brought the mission to an end. So each team

would work four long days journer's

first

a

week, with two days

off.

On

member

sol

So-

1,

day on Mars, the entire rover team would be present. There

was no way to

force people to

go home.

No one was going to miss the cul-

mination of years of work.

* How was the rover team going to get ready for landed operations? We were going to practice. The Pathfinder project had a plan to do Operations Readiness Tests, or ORTs. which were a cross between rehearsals and war

games. But they had not scheduled enough of them to

ORT involving the So

1

proposed

until landing.

rover wasn't until April, three

we conduct our own Rover ORTs That sounded

like a lot

suit

months before

We

wanted

people for more than one job so that we'd have some

training

sick at a critical

enough people

only get about four July

4.

moment. Between

to cover seven days a

tries at his

that

first

landing!

— RORTs—once

of testing (too much, to

those on the team), but that was deceptive.

somebody got

me. Their

a

week

many

of

to cross-train

flexibility in case

and the necessity of

week, each engineer would

or her job before having to do

it

for real

on

SOJOURNER

238

By early February, to cover

all

it

had become obvious that the team was just too thin

of the engineering positions during surface operations. The

overall mission operations process

would go on

pretty

much twenty-four

hours a day and some of us were going to have to work very long

While we were

how good

all

willing to stick

we would

a job

do.

it

out

shifts.

necessary; the big question

if

My greatest

was

concern was that the uplink

planners would be too fatigued after working twelve hours to pick up on their

own mistakes. We needed someone who could come in fresh and re-

view the commands

for correctness.

We needed more people.

Matt Wallace, from the rover power subsys-

tem, seemed to have just the right temperament to be a Rover Coordinator.

He was now

steal

him back?

dealing exclusively with lander also

I

power

issues

—could we

thought of Sharon Laubach. She was a Caltech

Ph.D. student in robotics, working on a rover research task at JPL. She had

gotten that job by hanging around the MarsYard helping out until she had

wrangled

a position

doing her dissertation research on the newest Rocky

on

several

willing.

Maybe

microrover. Laubach had offered to support Pathfinder rover occasions. She

we

was

bright, she

knew

rovers,

and she was

could train her to review sequences in time? I

gave Laubach a description of the rover's

set to study,

command sequences had generated over the previous months of testing. I didn't have much time, so I put her in a sink-

along with a set of several

command

or-swim position.

commanding

I

I

was

skeptical that

the rover in the few

Laubach studied the

materials, she

anyone could learn the

of

intricacies

months remaining before

landing.

As

would formulate questions, then come

me for specific answers. For a "final exam" presented her with a complex rover command sequence had written, and asked that she find

back to

I

I

the

bug

in

it.

It

was

a particularly difficult

something I had put in by mistake;

instead,

it

bug

to locate, for

I

invited Rick

Welch

was not

was something I had left out.

Laubach found the error on her own. She would do just Stone and

it

to join the team.

fine!

Welch had been

He had

taken

over as task manager on the Hazbot hazardous response robot job

when

working on rovers of one type or another

Stone got pulled into Pathfinder.

He had

at JPL for years.

since

microrover research tasks that had grown up to

moved on fill

to the

the gap

when

Rocky the

7

first

So What Are You Going to Do for the Next Six Months?

239

group of rover researchers joined the Sojourner team. Welch was an cellent systems engineer

He

also

had

sometimes vexing sense of humor. One of

a

ex-

with a background in mechanical engineering. his favorite pas-

them

times was starting unfounded rumors and watching

circulate.

When we met with Welch, he was low-key but agreed to become part

We focused on how to get him involved while allow-

of rover operations. ing

him

to

was more

fulfill

his

excited.

drive a rover

commitments on

As he commented much

that.

Come

One day

in

don t think so'?"

I

down an

going to turn

me to He sti-

"You guys ask

later:

on Mars. And I'm going to say 'No.

fled a laugh. "Yeah. Right. I'm

Welch

his current task. Privately,

opportunity

like

on!"

team

the rover downlink

cubicle, Allen Sirota

took

me

aside.

He warned me that was pushing people too hard on operations training. I

Some of

the

team members were getting pissed

they did wasn't good enough for me.

was giving them. For the attitude that

sponsible for

past three years,

he owned the its

rover, that

success as he was.

confront the possibility that in myself:

During development,

operations,

I still

I

Now,

acted as

if

I

Sirota

would

team

to take

summed up

on

Sirota's

on the team was

few words forced

a similar attitude.

I

I

as re-

me

to

could see

had been the operations "voice

I

it

in the

when the entire team was focused solely

was

its

was the impression

that

else

fighting the

too responsible for the success of the mission. the rest of the

because whatever

had resented Howard Eisen s

no one

was guilty of

I

wilderness" for so long that today,

on

Or at least

off,

war

And

alone.

that left

I

was

feeling

no room

for

own responsibility.

his advice to

me

in

two words: "Lighten

up."

I

struggle with that suggestion for months.

The Rover Operations Readiness Tests continued. to maintain the once-a-week schedule

we were

testing Marie Curie

all

the mechanics of running the

I

had hoped

the time. At

tests:

the evolving telemetry displays,

who

We would never manage

first,

for.

the

Even

it felt

as if

team stumbled over

was doing what,

what information

so,

to

how to interpret

communicate

be-

SOJOURNER

240

tween the downlink engineers and the uplink team. But the rover team learned quickly.

Even erations,

as the

team members were adjusting

we were

having trouble getting the rover to

Curie was just not doing well

we commanded

were followed by

far

sand,

it

was

its

we were

going where

targets.

asking

Or it might turn

twenty-five.

it

off course as

it

might

would

drift

primary means of knowing

inevitable slippage of

was not performing

"drifting." It

it

went.

rate sensor, the rover's

had turned despite the

Marie

to go. If

And if that turn

a long traverse of several meters, the rover

and farther

The onboard turn

how

at

roles for op-

the rover to turn twenty degrees to the right,

turn about twenty degrees.

farther

new

to their

consistently.

its

wheels in

soft

We had seen the

problem before, during our abbreviated driving tests with both Sojourner and Marie Curie prior to launch. In the rush to time to fully understand the rate sensor

seemed worse than Marie

delivery, there

drift.

Curie's, so the units

had been no

Sojourner's sensor had

had been swapped before

the flight rover shipped to the Cape. Other than that attempt to mitigate the problem,

we had been

trapped by one of the primary constraints of

We had been forced to live with a known hardware problem, for there was no time to correct We thought flight projects:

The

planets don't wait.

it.

now that we knew the

culprit: noise. Just as electronic noise

bane of the APXS, noise

in the

power supply

had been the

lines feeding the rate sensor

we could do was look for software fixes and operations workarounds to somehow compensate for it. This was the nature of one-of-a-kind flight systems versus mass-produced products. An

was corrupting its output.

All

automobile manufacturer would build a prototype, learn about

and then correct them

in the production

model. In a

its

quirks,

flight project,

we

flew the quirks.

Most of the RORTs were conducted in the Building 107 sandbox, with

commands coming from Rover Control Workstation in the flight rover control room on the other side of JPL. But for a few days in March, Marie Curie was not in the sandbox, but instead in the Mojave Desert near an extinct volcanic crater called

that

Donna

Shirley

Amboy We were

had been

insisting

Weather had aborted the

on

field tests

finally

for the last

doing the

two

field tests

years.

twice before. High winds in the

So What Are You Going to Do for the Next Six Months?

Amboy

desert along the approach route and at to the field site dangerous. Art

Crater could

The motor homes would

field

command

ned

the satellite

Microrover

getting

two days of calm conditions

NASA flatbed truck.

before sending out the caravan of two rented RYs. a a van.

make

Thompson had been checking the weather

reports daily waiting for a forecast of at least

and

241

serve as sleeping quarters and the

center with workstations and equipment.

The truck

ear-

communications dish antenna and the old Mars Science

MSM

simulated lander.

The

cophagus."' rode in the van.

Thompson. Allen

Mane

field

Curie, safely boxed in

its

"sar-

team consisted of four people: Art

George Alahuzos. and Jim Lloyd. Thompson

Sirota.

gave the go-ahead early on a Sunday morning. Their destination was an

"We took Highway

isolated spot.

solutely

The

no

more

to

RYs were sinking

axles.

hour away

The

It

would

— to

field

and got off where

take a

free the

into the sand.

buned one of

they immediately

site,

its

the

Forty- east,

drove for about an hour out to

this

Once they reached

motor homes

first

test site set

stereo images taken

up by

first

traverse for early

For the

by the

Then the

sand up

—over an

for the return trip.

set

early afternoon

on Sunday,

up the antenna dish and

tablished the satellite link to JPL in about thirty- minutes.

light in the late afternoon.

in the

tow truck from the nearest town

motor home

team had the

the

sooner than they had expected. They had

get the

there's ab-

dead volcano.'"

convoy drove up the undeveloped road toward the

farther the small

crater, the

the

civilization,

MSM

es-

The plan was

to

lander before they lost the

rover drivers could plan Marie Curie's

Monday morning.

field test, the rest

of the rover team stayed

at

JPL. .After

all.

the

point of the test was to operate the rover as realisticallv as possible, with the operations

come back

team relying

in telemetry

solely

on the same kind of data

that

would

from Mars. So the JPL crew would look only

at

images and engineering data generated onboard Marie Cune. It

was only

much testing

a test,

but the pressure was on.

as possible

the satellite dish over.

We

So

we

to get in as

while the weather held. Fierce winds could blow

The sand those winds would carry could scour

outside of Marie Curie, scratch camera lenses, and electronics.

needed

tried to

into each day of testing.

cram three or four

sols'

work

its

way

the

into the

worth of operations

The command sequences were dominated by

SOJOURNER

242

and APXS placement, without any time-consuming experiments

traverses

down. The hope was to design the next sequence and up-

to slow things link

it

sults,

within two hours,

and

start

let

the rover operate for an hour or so, see the re-

planning another sequence immediately. We'd repeat the

process until the lander images were too dark to see what was going on.

The

team spent most of

field

its

time waiting.

would always be there keeping watch over Marie of the rover cameras.

When the

rover

was

One

or two of

them

Curie, just out of view

team mem-

in motion, a field

ber would pick up Marie Curie's power cable and walk along behind, making sure that the cable didn't get tangled at a time, the rover did

team

at JPL would be

building

very

little.

But for hours

During these periods, the operations

furiously arguing over the best course of action, or

.

.

.

and plan dinner.

One of Thompson's most the meals.

They

excellent chef.

I

certainly

would

done." For their efforts

vivid

memories of the

had plenty of time

say that

for cooking.

was some of

Henry Stone had

field test

would be

"George

is

an

the best eating we'd ever

"bribed" the field

team with

He had given them almost five pounds of rock shrimp he'd brought

back from Florida field

rocks.

command sequences. The field team had nothing to do but listen

to the JPL chatter

food:

among the

in

an

ice chest after the Pathfinder launch.

team gladly accepted the shrimp, they

off point for their

treated

own culinary plans. "We

it

While the

as only the jumping-

ate really well."

The rover testing was not so memorable. There were few rocks in the lander panorama big enough to be good APXS targets. The best target would have been

a large, flat-sided boulder, the

"the broadside of a barn," something so big

we

Martian equivalent of

couldn't miss

it.

But there

weren't any of those. Instead, the rocks were so small that Marie Curie

might accidentally drive

right over them.

And that meant the APXS would

have to be precisely placed to make contact. The JPL crew picked a First

we had

to drive the rover to the vicinity of the rock. Brian

target.

Cooper

planned the traverses, studying the scene through his stereo goggles. sent the sequence.

overshot. rock.

We

When

the

new images came

We

back, Marie Curie had

struggled through the next sequences, zeroing in on our

When we

thought the rover was properly aligned,

we commanded

So What Are You Going to Do

for the

Next Six Months?

Marie Curie to back up a few inches, then activate the

Mechanism

to finally place the sensor

The new images were away from its it

personally.

haps the

target, I

head onto the rock.

The APXS was only inches but we had missed again. I was frustrated, and taking hit the

rock with one more sequence. Per-

MSM lander cameras had been knocked out of calibration dur-

rover turns. But

site.

Perhaps the rate sensor

we were

drift

was throwing off the

out of time.

By noon Wednesday

a

new weather

team came home. The small group

that

front

was moving

The

in.

field

had braved the desert boredom

now been jokingly branded "The Amboys." The mixed

boy

APXS Deployment

a disappointment.

knew we could

ing transport to the

had

243

results

from our attempts to get

to the target rock at

Crater, together with similar experiences during earlier

RORTs,

left

the

team with

a

looming uncertainty: Would we

to reach rocks in a day or so

journer's mission a success?

We would keep practicing.

on Mars? Would we be

sandbox

really

able to

Am-

be able

make

So-

TWENTY-TWO

TESTING, TESTING

Friction between for

most of the

the rover

team and the project team had

lifetime of the project.

persisted

To many of the rover team

members, the dichotomy was simple: "We" were the rover and "they" were the lander.

The

difficulties

between the two teams were most evident during the

integration and test phase of the project. Everybody

was working

hard.

Everyone was under pressure. To the rover team, the lander team seemed

To the lander people,

unresponsive.

manding. The lander had

its

the rover group

own problems

was arrogant and

de-

to deal with; they'd get

around to things the rover needed, but the rover was never willing to compromise.

The

The

rover guys always wanted their tests run now!

rover and lander engineers saw themselves as

moved

teams. But after launch, as the Pathfinder spacecraft

toward

its

der system engineer Dave Gruel recalled

was

it

we had

'My

a

stuff

working. Your it all

inexorably

rendezvous with Mars, that perception had to change. As later,

team were the ORTs [Operations Readiness realized

two separate

stuff's

doesn't!'

really

made

us one

we No more

Tests]. After the first one,

common enemy We were works and yours

"What

lan-

all flailing

Now

it

around.

was 'My

stuff's

not

not working. We've got to come together and make

work, or we're in deep trouble.'"

Testing, Testing

.

245

.

.

The "common enemy" paradigm had come design and development of the rover and levels. Conflicts

maraderie

into play throughout the

Pathfinder, operating at several

between rover subsystems tended

among

to strengthen the ca-

the individuals within a subsystem. During the pro-

tracted effort to solve the "Principal Investigator

from

APXS

noise problem,

Hell,"

became

Tom Economou,

the

the bigger enemy, and thereby

helped the rover team to bond more tightly together. The rover team was further unified by our shared perception that the lander agenda threat-

ened the ultimate goal of

a successful rover mission.

And

then, with the

prospect of mission failure looming ahead, as simulated in the Operations

Readiness Tests, the rover and project teams finally began to meld together.

The ORTs were full-up simulations of on-Mars

make very

those simulations

seem

own little piece of Mars:

operations. In order to

as real as possible, the project

needed

its

the "sandbox." Just as the Building 107 sand-

box had been the rover team's arena

for testing

and improving the

rover's

performance, the Pathfinder sandbox would validate overall surface operations.

The

project's

sandbox was a sealed room on the second floor of

the Space Flight Operations Center. the floor. There

was

people came and went; sets

anteroom

a small this

White sand

real

inches deep covered

through which

in the corner

required anyone entering to pass through two

of doors. There were two reasons for

from the limited but

six

this: first,

to protect people

hazard of the rover's laser stripe projectors, and

second, to prevent the migration of sand from the sandbox to the rest of the second floor.

The mission operations

area

was

filled

with Macintosh

and Sun workstations, and sand was the bane of computer hard

The engineering model of

the lander

was placed

drives.

in the sandbox.

Track

lighting allowed for sufficient illumination to operate the

IMP and

cameras, but not nearly enough to simulate Mars sunlight.

A photomural

of one of the Viking landing testbed

dow

sites

was located on the other

covered part of two walls. The

side of the sandbox's

north wall.

rover

flight

A win-

in the center of the wall allowed the engineers in the testbed to ob-

serve the activities in the sandbox. There

were

also

two windows on

the

south wall of the sandbox, so people in the adjacent hallway could look in.

During

tests,

blinds covered those

windows. The doors to both the

test-

SOJOURNER

246

bed and the sandbox were sealed by

cipher-locks.

Only those people with

in. The point, during ORTs at least, was make the sandbox as far away as Mars. The operations teams would only know what they could discern from the telemetry stream coming back from the lander and rover. The teams would have to learn how to

the right combination could get to

command

the mission, assess the health of the spacecraft, and recover

from anomalies with only the incomplete information Pathfinder could send home.

As part of giving the operations teams much needed experience operating the lander

and rover the way they would have to on Mars, the

ness tests forced the teams to live the after Pathfinder landed.

We

work

shifts that

would be

readi-

in place

would be generating new commands while

the lander and rover "slept" during the Martian night. Only that

would commands be ready the early

when

it

so

direct line-of-sight to the lander.

happened

be about seven hours

that

on July

earlier

4,

the time at the landing

which would be about 10 a.m. Planner for sol as late as

I

the midnight

main awake

2.

in Pasadena.

I

Time (MLST),

was going to be the Sequence I

shift.

I

until the

would

shift

my personal biological clock to

rent four or five videos, then struggle to re-

sun came up.

The ORTs were always focused on landing and the

some

first

few

subset of that time period. The

early surface mission required the execution of a

to put the lander

would

Two days before each ORT, would force myself to stay

could in an attempt to

face mission, or

site

than Pacific Daylight Time on Earth: The

landing would take place about 3 a.m. Mars Local Solar

up

way

Earth rose in

morning Martian sky and the big antennas of the Deep Space

Network had

And

to transmit to the lander

and rover

sols

of the sur-

activities

of the

complex choreography

in a stable state so that they

would survive and

achieve the mission objectives.

Rob Manning and try,

part of the lander operations

team simulated En-

Descent, and Landing (EDL) over and over again. In

was not much

in the

way of EDL

"operations" for the

some team

sense there to train for:

the lander acrobatics necessary to drop from interplanetary speeds to a

.

Testing, Testing

stable perch

on the

surface of

entire

take any message

.

24?

.

Mars would happen too

EDL

intervention from Earth.

would

.

would begin and end

from the lander to

fast for

human

time than

in less

DSN.

arrive at the

any

it

In fact, the

Entry Descent, and Landing process would be activated by

a single

command from the Pathfinder operators, called "Do_EDL." And this command merely authorized the lander to begin running the onboard software that would do the right things when the time came. Despite

of

all

that, there

EDL. There might be

was great value

subtle bugs

hidden

still

in

running simulations of

in the software. Slight

anom-

or perturbations to the lander's performance might affect the

alies

ware's execution in unexpected ways, perhaps causing

Manning and there

was

his

team could

find

any such

to

fail.

If

sensitivities in simulation,

time to correct them, and upload

still

EDL

soft-

new

software to the

Pathfinder spacecraft already en route to Mars.

EDL would also make the next part of the ORT more The EDL telemetry data would be some of the first information

Simulating realistic.

received from the spacecraft after a successful landing.

The operations teams tailed sol ect.

1

ORTs would attempt

start to the surface mission, the series

would need

to occur

in rapid fire.

age to the spacecraft due to the landing to

to follow the de-

scenario that had evolved over the prior three years of the proj-

For a successful

scenario

in the

happen on

sol

1

As long

of events in the

most of the

itself,

was no dam-

as there

steps that

had

could be handled by a set of pre-canned sequences.

The lander and rover operations teams on Earth would just be required to

make

a

number of go /no-go

command to actimemory on team would normally command Path-

decisions, then send the

vate the appropriate sequence already stored in the lander's

Mars. After sol

1,

the operations

finder only once per sol.

While

mission (and represented a prior mission),

Once signal

on

it

its

its

sufficient for

most of the

much faster command turnaround than in any

was on the ground,

it

for sol

1

would transmit

a

confirming

thumb-sized antenna. That signal would not be heard

Earth, unless the lander

that

would be

would not be good enough

the lander

through

this

happened

to

tiny antenna pointed at the sky.

could do for the moment.

It

had other

come

to rest base-petal

down, so

But that was the best the lander

tasks to

perform before

it

would be

SOJOURNER

248

able to send a

more

certain

message home. The airbags would need to be

vented and retracted out of the way; otherwise they would remain draped over everything near the lander, precluding close-up images of the local terrain

by the IMP camera, and impeding rover

tors in each lander petal

would

egress.

So retraction mo-

operate, reeling in the Kevlar cords that

ran through the airbags onto take-up spools, dragging the bags across the ground, and compressing

each

them more

tightly against the backside of

petal.

Airbags safely stowed, Pathfinder would proceed to open

its

petals,

the inside surfaces of which were mostly covered with solar cells to replenish the depleted lander battery.

the other three were attached,

was

Only the

and instruments. In the

lander's computer, radio, antennas,

opening, the lander would set tation.

itself upright,

Onboard sensors would

central base petal, to

free of solar cells; instead

tell

no matter

the lander

how

it

its

sat

it

which

carried the

act of petal

original orien-

on the

surface.

petal was down would open first, until Pathfinder fell onto its Then all three outer petals would open fully, leaving Pathfinder

Whichever base petal.

ready for

its first

Martian sunrise.

With the petals open, the lander's low-gain antenna would now be upright a

and exposed to the heavens. Less than four hours

little

first

over an hour after sunrise, Pathfinder

message

the message

likely to

after landing,

and

would begin transmitting the

be detected on Earth by the DSN. The contents of

would be engineering data about EDL, along with

status

about the lander and rover subsystems. But the meaning would be "I'M

HERE!" The

operations teams

would now have some work

viewing the downlinked telemetry. they would then uplink

commands

with the nominal mission plan. era

would begin taking

If

Still

no

significant

to do, quickly re-

anomalies were found,

authorizing the lander to continue in

its

stereo pictures.

stowed position, the IMP cam-

When

combined, these images

would form the "insurance panorama," so named because that at least

some photographic record of

to Earth in case the

IMP

failed

soon

the landing

after landing.

potential glitch that could cause the loss of the

deployment. The

IMP

site

(One

it

would ensure

would make

it

clearly identified

IMP would be

a failed

mast

mast was a spring-driven device with significant

Testing, Testing

energy stored in

two and

it,

structure, just

But

the lander

if

mast was

ment body

the

249

.

released,

it

was only about

snapped upward to

it

happened

on

to land

IMP mast might

a

a

to put the jester

a rigid

camera steady for picture-taking.

rock that

tilted its

base

deploy unevenly and

version of a jack-in-the-box.

more than over.

fall

The

the deploy-

If

IMP head would end up hanging upside-down,

in the breeze.

its full

Locked in its deployed state, the mast was

feet.

like a sophisticated

failed,

bobbing

when

tall;

what you wanted to hold

thirty degrees, the

.

similar to the rover ramps. Stowed,

a half inches

height of over three

.

useless,

There would be no second chances. There was noback

in the

With the sun high enough

box

in the

to try again.)

Martian

sky,

the

IMP camera would

begin searching for that sun. Software running on the lander's computer

would look for

a set of bright pixels in the images.

ware used the position of the sun in the

sky,

When found,

together with the

the soft-

known time

of day, to uniquely determine the direction of north. (Mars has no magnetic field, so a

clock to

know

the time and

face, the lander

Earth.

magnetic compass would not function

could

now

knowing

track the

The lander would point

its

orientation

there.)

Using

on the Martian

most important object

in

its

its

sur-

sky: the

high-gain antenna and at the proper

its

time begin transmitting data, including images,

at a far greater rate

than

the low-gain antenna could ever achieve. If everything were operating perfectly, the

hours

the downlinked photos

to decide if

team determined

rises that

that

commands would go up

it

later,

safe to

The

new

the

set

of

APXS sensor the APXS at the mo-

position of the

which might

strike

The next rover command sequence would cause ADM fully, taking the APXS out of harm's way

unrolled.

the

deploy the ramps, a

If

More pvros would fire, and the and its APXS sensor head would be cut loose.

yet be released:

the rover to retract the

the images, in stereo,

might block the unrolling ramps.

was

close to the rear ramp,

ment the ramp

Moments

it

the rover

The Rover

to the spacecraft.

tie-downs of both the rover

The ramps could not

safely released.

members would examine

looking for any rocks or

head put

would be those needed by

one or both ramps could be

Driver and other team

rover

six

after landing.

Among team

transmission would start at 9 a.m. Mars local time, about

ramps would

deploy.

SOJOURNER

250

The

third rover

stand up.

The

rover

command would

sequence of

drive

maximum ground

height and

could then take the

its

would

1

tell

the rover to

rear wheels forward, rising to

From

clearance.

of images with

first set

sol

The IMP would take more pictures,

its

this

its full

this vantage, the rover

own

cameras.

time to confirm that the rover

ramps were properly in place, ready to be driven on. After the unavoidable time delay caused by Mars's great distance, the images would arrive on Earth, and the rover operations

"Which way do we

tions,

go

at the

team would scramble

drive?

Forward or back? Which way should we

bottom of the ramp?" From

this point on,

canned rover sequence would do. The terrain

would be

totally

unknown

to answer the ques-

no completely

at the

pre-

end of the ramp

IMP images trickled in. The now recommend the first

until those first

Experiment Operations Working Group would

APXS

target, either soil or a

the rover's

convenient rock. The driver would then plan

first traverse.

Dozens of pyrotechnic to the surface: cruise stage

ment, airbag

inflation,

firings

were necessary to get Pathfinder

safely

and heatshield separations, parachute deploy-

and more. Each of these pyrotechnics was

gered using energy from the lander battery, and the combination of

them would have 1,

left

the lander battery

the battery severely depleted. As the sun rose

would

likely

have

less

than half of

its

trigall

on

of

sol

capacity re-

maining. Sunlight on the solar panels would begin recharging the battery,

but as long as the rover

sat

on one

would be shadowed, making the der's total solar

power would

petal,

most of

petal useless as a

therefore be

down

that petal's solar cells

power

a third.

source.

The

lan-

The charging of

the battery might not keep pace with the drain from the various lander

The lander operators wanted the rover off the petal as soon as possible, to bring the solar arrays up to full strength. Following the sol 1 plan, the rover team obliges. The new sequence is uplinked, and the rover drives. The rover rolls off the ramp and onto subsystems.

Mars.

The IMP

rover.

When

rover had done tions

takes an end-of-sol photo of the predicted location of the

the images finally arrive its

job.

on

Earth, they confirm that the

Toward the end of the

first sol,

the lander opera-

team commands the lander to deploy the IMP mast to

The lander and

rover

would now be ready

to

do

its full

science.

height.

Testing, Testing

And any ORT

.

.

251

.

that reached this point in the plan

by the end of

sol

1

would be going smoothly indeed.

Dave Gruel was the "Gremlin," the engineer responsible for injecting prob-

He would

lems into the ORTs. ating dunes

rearrange the terrain in the sandbox, cre-

and rock distributions

challenge the operations teams.

he saw

as

fit,

positioning the lander to

The job required

creativity, a bit

of a mis-

chievous attitude, and the ability to keep a secret. Gruel was good at

The

final

tempers often flared ficult.

at the

person responsible for making their lives so

Richard

full-scale surface

first

Cook had told Gruel

to

operations ORT, Mission Manager

make

the terrain difficult.

He

dose of reality for the team. Gruel obliged.

large dune,

was the

The problem for the

system.

rover

largest such feature

would allow the Marie Curie rover

safe egress.

team chose the forward

sat

direction,

Gruel saw to

ramp would

gle,

beyond what the rover had been designed

doable.

rest if deployed.

To make it

sure the

that a deployed

it

was way too steep

to

to negotiate.

go

off the back.

"I I

expected

purposely

horrific."

But things did not work out according to the Gremlin's plan. the

first

he saw

Howard high dune and trough next to it made

images came back

it,

the

impossible.

if

slope steeply downhill, at a forty-five- or fifty-degree an-

the rover guys to say tilt

in the solar

on Olympus Mons and

act as a bridge.

rear

the

after

Gruel designed the

where the forward ramp would

That meant that the ramp would have to

made

known

was challenging but

trough between where the lander

a

the next rise in the sand

rover

would be the

team would be deciding which ramp,

terrain carefully, to create a situation that

There was

It

placed the lander on a

which he and Rob Manning christened "Olympus Mons"

the Martian volcano that

any,

dif-

Gruel could handle that part of the job too.

For the

first

it.

job qualification for a Gremlin was a thick skin, since people's

It

was

after "landing,"

a totally unrealistic situation!

When

Eisen was furious. As rover drive-off nearly

He was convinced that He went to complain

such terrain would never actually happen on Mars. to Gruel.

When

the forward

Brian Cooper estimated from

ramp would

contact the ground,

it

IMP

stereo images

was obvious

to

him

where that

it

SOJOURNER

252

was going

to be unacceptable.

maneuver once

it

drove

off.

There would be no room

for the rover to

Therefore, the rover had to go

down the

rear

ramp! The rover team requested that only the rear ramp be deployed.

The few people who were allowed like

Dave Gruel and Jake

Matijevic,

into the sandbox during the

were precluded from participating

the key operations decisions taking place.

because they could only

sit

by

ORT, in

was very frustrating to them,

It

as the rover operations

based on the telemetry data alone, and could not so

team made choices

much

as exhibit a re-

vealing facial expression.

For the next three days of the together to

command the

ORT the rover and lander teams worked

lander petal deployment motors to

der into a better position for deploying the rear ramp.

were participating

tists

ORT

reached Mars. But as the

proceeded,

would be going anywhere, and from

the test looked like erations

team was

it

A group

the lan-

of scien-

in the test, waiting for the chance to practice

directing the rover to science targets as they

rover

tilt

was going to be

satisfied

it

would when Sojourner

seemed

unlikely that the

a science training perspective,

a washout. But, finally, the rover op-

and asked the lander team to deploy the rear

ramp.

The

project could not afford a perfect duplicate of the flight lander to

complement the

flightlike

Marie Curie

rover. Instead, the

achieved flightlike capability piecemeal.

The

lander's

sandbox lander

computer was

actu-

located in the next room, in the testbed area, with cables running

ally

through the wall to the lander hardware to deploy the

in the sandbox.

IMP camera, someone would bolt a mast of the appropriate

height to the lander body, and the camera

"Deployment" of setting

When it was time

a rover egress

ramp

itself to

the top of the mast.

consisted of an engineer manually

down and aligning the ramp with one end on the

the other

on

lander petal and

sand.

Although Matijevic was the Rover Manager, during the

on the over

role of carrying cables,

its

making

ORT he took

sure the Marie Curie did not drive

own power lines, and deploying ramps. When he heard that it was

time to deploy the ramps, he placed both engineering model ramps in place,

one forward and one

rear.

He knew from his direct view of the rear

Testing, Testing ...

ramp

IMP camera had

ramp was

way

the

to go.

He

to be deployed.

ward ramp

had not intended

team another

was the only sol 3

data

forward

for the

over.

The IMP panoramas

in place. For better or worse, the rover operators

take the rover

APXS

the

later, after

to be taken again, without the erroneously positioned for-

to live with their decision. Relying only

On

was only

rushed over to Rover Control to intercept the

images before the operators looked them

would have

It

a steep angle.

taken the stereo rover deployment panoramas, that

Matijevic discovered the rover operators

tion

down such

that the rover could never drive safely

Clearly the forward

ramp

253

telemetry,

it

would

day to conclude that the forward

full

finally

direc-

down

drove

the forward ramp.

Some

ORT

and some images were captured. The

collected,

ended ingloriously on

sol 5.

At the end of the ORT, Dave Gruel hated the scientists.

would have

viable solution.

Marie Curie

was

on appropriate

He had done

his job, the rover

team, and

the best he could to challenge the operations

team, which was exactly what he had been asked to do. But everyone had

come screaming to him to complain. The rover team had complained that he had made their job impossible. The science team had pointed out that Gruel was not a geologist, and that the terrain features he had created could not exist in nature.

He had

gotten an earful.

From

the operations

teams' point of view, nothing had gone right. Looking back at

considered that

ORT

you could not imagine

"the biggest mess,

it,

Gruel

a test go-

ing worse." Yet,

from another

perspective, the

ORT

had proven that we were not yet ready team, test

procedures, nor

its

all

had revealed problems

when the

of

its

early,

actual landing occurred.

had been

a

major

software tools were yet mature.

so that they It

also

and launched.

that told

The

would not happen again

marked the end of the hardware had neces-

been on the back burner during the rush to get the spacecraft

tested,

It

for Mars. Neither the operations

bias of the Pathfinder project. Surface operations preparation sarily

success:

built,

Now everyone on the project had had an experience

them operations would not just take

most Pathfinder project personnel, the

ORT

care of themselves.

had been the

first

And

for

exposure

SOJOURNER

254

to the steps involved in operating a rover.

The Sojourner and

Pathfinder

teams began to grasp the complexities of operating their respective spacecraft.

The ORTs slowly improved.

The teams calmed down between the

tests,

focused on what they had learned. Dave Gruel continued to invent

and

new

problems for both lander and rover teams to overcome. The teams even

became To

less belligerent

set

toward the Gremlin.

up the sandbox, Gruel would come

before an ORT,

when no one

in

on the Sunday afternoon

was around. He'd go

else

with a shovel and rearrange things.

If

he didn't

into the sandbox

like the results, he'd re-

arrange the terrain again. While the rest of his job was getting better, he still

hated setting up the sandbox.

It

was exhausting

physical work, and

shoveling kicked up clouds of fine dust that he had to try to avoid breathing. Partly to

make

the setup task less distasteful, and partly because his

"victims" could use a break, Gruel started to place foreign objects into the

sandbox before each

test.

He

couldn't

make

the

ORTs any

easier,

but he

humor into the tests. "ORTs were a highIf there was something we could do to lighten things up, that

could find ways to inject some stress time.

was

a goal." For

one ORT, Gruel positioned

a plastic skeleton

facing the lander and gesturing toward the IMP. In another

sand

pail

and

plastic rocket toy

were

visible.

and the operations teams were

the downlinked

IMP images

The

geologists studying the

test, a child's

in the habit of, searching

for the Gremlin's signs, all

a rock,

After a few ORTs, the partic-

ipating scientists

whether they would recognize

on

and wondering

of them.

initial

"mission success panorama" dur-

ing one test indeed discovered something unusual. Additional multispectral

imagery confirmed the

find:

a potted plant.

At the Experiment

Operations Working Group meeting, the science team took the

ommending

bait, rec-

that the rover get a close look at this evidence of potential

The rover team was less sanguine about sending the rover to investigate. The leafy green life form tentatively identified was in the far corner of the sandbox, and would reas a philodendron quire a traverse of several meters to reach. The rover never did get to the Martian vegetable

life.

—

—

Testing, Testing

plant before the completion of the

rover traverse

up

..

255

.

ORT, but we did manage the longest

to that time.

For another of the ORTs, Gruel draped deflated airbag material over the rover petal, blocking the deployment of both ramps.

came up with

a

commands

lift

to

scheme

to solve the problem.

The

lander team

They put together

a set of

the offending petal off the sand, operate the airbag re-

traction motors, drawing the airbag material

more

tightly against the out-

lower the petal back onto the surface. After the

side of the petal, then

IMP images confirmed

sequence went to the testbed,

that the airbags

were out of the way, and the rover mission proceeded.

What if the lander's battery failed after landing, allowing the lander to operate only when the sun was high in the sky? What if the high-gain antenna never locked onto the Earth, and the entire mission had to rely only

on the low-gain antenna? The ORTs gave the operations team experience dealing with contingency scenarios like these.

came up with new commands, and

the engineers monitoring

and thermal conditions learned the use of

files

refined

them when

never enough

test

necessary.

ORTs

of saying,

did not go perfectly, and there

was

"No plan

The rover team continued to do project's

ORTs. There was

of a full-up point

Bill

ORT

a lot

its

the expertise they

all

flexible.

This was key As

will survive sol 1."

own RORTs, interspersed with the

we were learning without the

constraints

involving the entire project operations team. At one

Dias, the project's lead surface operations planner, joked that

should conduct Surface Normal

In

power proand

time for the engineers to gain

Cook was fond

builders

their software tools,

wanted. But the operations teams learned to be Richard

The sequence

early May, the

ORTs

Hot Wheels toy

we

We never did.

or "SNORTs."

rover hit the stores.

Howard

Eisen had

worked nights and weekends with the Mattel designers over the period of a year to design a tiny

in short supply, I

we

Friday, figuring

they had any in stock. next store didn't

the sets were initially

scrambled to find one.

all

went home one

model of Sojourner. Though

The

first

I

would

call a

few toy stores to see

store said simply no.

know if they were

in stock, so she

The operator

had

if

at the

me wait while

she

SOJOURNER

256

me

connected wanted,

was

I

with someone on the told,

we had

"Oh, yeah,

When

floor.

a

few come

only a few Three cases of three each. They Ve the

phone had no idea when more would come

in.

might look. There was a pause. "Well, you know, aside for myself.

one."

I

Along the way,

to scalp the rover. I

them, so

really collect

I

sold."

asked where else

do have one that

could

I

The guy on

let

set

I

you have

I

that

name and headed out to the store, which was all of ten min-

got his

utes away.

don t

I

I

I

on Wednesday, but

been

all

explained what

I

in

had with me.

I

I

considered that he was probably going to try

wondered how much I would pay, and how much cash

When

I

got to the store,

I

He

quickly located the salesman.

took me to the back and brought out the box. Inside was the model of Sojourner, flanked

by the Pathfinder lander and the

what you're looking

for?"

he asked. "That's

cruise stage. "Is that

Instead of asking for

it."

money, he simply turned and walked away, disappearing down aisle.

A

I

went up

look on his I

to the register

few days

later,

I

and

paid: five dollars plus tax.

passed by Matt Wallace's

He looked over at me.

face.

"I

office.

just got a call

He had

from

mentioned to her a few days ago that the Hot Wheels were

they were hard to

come

by. She's like a bulldog.

the toy stores in the Washington,

to do?

told her, 'Buy

I

it!

I

saw

her.

my mother.

out,

and that

D.C., area. She just called to say she

Ship

it.

What

all

had

did

I

it!'"

My secretary would report to me her progress every time

a funny

She must have called

located a case that she thinks has at least twelve rovers in

want her

a store

She was clearly on her

own

on

locating the rovers

crusade. "There's

none

available in Arizona!" she announced one day.

Eventually, just before landing day, a large rived

and were

distributed,

"payment" from Mattel

number of

toy rovers

ar-

one to each member of the Pathfinder team for

the

design assistance provided by the

Sojourner team.

The rover primary mission had always been stated to be one week, or seven

Martian

sols.

But most of the rover team thought the rover would keep

driving after that. landing,

and

Matt Wallace

we were

said that if the lander

able to get Sojourner stood

and rover survived

up and down the ramp,

Testing, Testing

"the rover will last for a

.

.

25?

.

good long time." He was betting the rover would

survive as long as the lander.

The them

most of the

attitude of

scientists

was very

different.

saw the rover as a nuisance, draining resources they

The

use themselves.

rover

would cut

into the available

bandwidth with the telemetry volume

down to

it

as gospel,

less bit available

lifetime of the rover

one week on the

after

would rather

generated: Every bit of data sent

They took the promised seven-day

and assumed that

of

communications

Earth to support rover operations would be one

for their science.

Many

surface, Sojourner

would be dead, and they could proceed with their science unencumbered. Most of the

science

also

new APXS

daily basis, getting a

RORTs and ORTs, hard.

to oversimplify rover opera-

the rover

reading each night.

team knew

From all of our APXS on target

that getting the

Tom Economou's viewpoint was that the rover team just wasn't

committed

to getting

APXS

data.

an ORT, he stood up and raised

team

seemed

imagining that the rover would just trundle from rock to rock on a

tions,

was

team

to get to this rock!" he

At one science planning meeting during

his fist into the

challenge the rover

air. "I

demanded in his raspy voice.

Project Scientist

Matt Golombek agreed that the rover team should go for broke and try to reach the chosen rock in one attempt. That attempt required a complex set

of maneuvers across several meters, which together with Marie

Curie's inherent dead reckoning error resulted in a final rover position a

long to

way from

do that

If it

again.

takes us

Golombek got the point: "We are never going are going to work within the capabilities of the rover.

the target.

We

two days

to get to a rock that's

what

it's

going to be. We're

not going to go ten meters to get to a rock. We're going to go a meter or two." But

would the

rest

of the science team understand?

* Just before the final

Rover

Scientist,

ORT, Hank Moore, now the

appointed

gave a talk he called "The Cussedness of Inanimate Ob-

jects." In his late sixties,

had been

officially

Moore was the

a planetary geologist

"old

working

man" of the

at the

science team.

United States Geological

Survey since the 1950s. In the mid-sixties he helped select the landing for the

He

sites

manned Apollo missions to the Moon, and gave the astronauts geo-

SOJOURNER

258

logical training before they got there.

But Mars had always been a special

place for Moore; in the 1970s, he identified sites for the Viking landers,

and continued analyzing the geology of Mars

widely

known

—

among

at least

touched

after the landers

down. His ongoing study of the planet eventually led him planetary geologists

to derive the

—Moore model of

rock distributions: This was a mathematical representation of the number of rocks of a given size you'd expect to find

(When

given location on Mars.

at a

rover designers at JPL designed test courses, or analyzed terrains

to determine

what

size a rover

needed to be to overcome

obstacles, they

used Moore's model.) In time, Moore retired from the Geological Survey.

But he came out of retirement to be the Rover Pathfinder.

erators

He seemed to be

always smiling.

whenever they got Marie Curie

to

Scientist

on Mars

And he praised the rover op-

do something

like

it

was

sup-

posed to do during ORTs. The impression you got was that he was having the time of his

life.

Moore's involvement in Viking twenty-two years before gave him a

unique perspective, and

his presentation

had

a

good turnout. His

basic

premise was that even rocks wouldn't do what you expect them to do.

warned the Pathfinder team smoothly

as they

push the rock sampler.

that rover operations

would not go

were assuming. During the Viking mission, "We

called

ICL

Computer Load) with

(for Initial

ICL did not move. Another

rock, Badger, rotated

the Lander 2 surface sampler. Badger finally yielded, but he his

He as

tried to

the Lander 2

and leaned on is still

there

on

haunches, ready to do battle with the Viking surface sampler." During

Viking,

it

had taken weeks

between when the

commands

a scientist finally

to

move a few rocks around. The

average delay

proposed an experiment on Viking and when

were sent

to

Mars had been twenty-two

days.

On

Pathfinder, the rover team was going to be turning around sequences

every

sol!

Moore

closed his talk with a plea to the assembled scientists for

patience during rover operations.

was

One of

the great desires of Pathfinder

to find "fresh" rock, free of dust, for the

against a dusty rock, the

APXS would

Demonstrating the

to analyze. Placed

probably determine the composi-

tion of the dust, instead of the rock under will take time.

APXS

it.

"Locating such rock surfaces

accessibility

take time. Accessing and positioning the

of such rock surfaces will

APXS

will take time." If So-

Testing, Testing

.

journer didn't get to that special rock in a

maybe

the next day, or Later,

Moore

back in the

the day after that

office

we

.

team who both had the

Tom Economou, who seemed

about his instrument,

it

get there

.

Henry Stone and I declared Hank He seemed to be the only person on the stature to

others and understood our problems. After the

with

would

day, well, she

shared,

the rover team's savior.

Pathfinder science

.

259

.

.

was good

to

be listened to by the

many months

of arguing

team

didn't care

to feel that the

rover

know that there was someone in our

corner.

ORT was our dress rehearsal for landing day and the

The seventh and

last

week of surface

operations that

would follow it. For the

the ORTs, the simulated mission

first

The

all

of

went smoothly. The lander experienced

few anomalies, and we got the rover down the ramp on planned.

time in

terrain of the landing site

was

sol

1,

just as

relatively flat; the rover

ma-

neuvered well in the sandbox and reached its designated targets. The rover

team reported Marie

Working Group meeting, and the

tions

next

Curie's progress each day in the Experiment Opera-

sol's

All in

rover

all, it

Of

was

course,

activities.

The

a successful

was

it

all

Pathfinder landing

was only

instill

its

APXS.

one-week mission.

ORT

this

ORT was to

rover took pictures and operated

planned that way. Richard

Gremlin to make

final

science teams set priorities for the

an easy one. a couple of

It

Cook had

asked the

was mid-June and the

real

weeks away. The purpose of the

confidence in the operations teams that

we were

ready for Mars. The teams were as trained as they would get. Dave Gruel inserted

no

significant

problems to vex the operations team. So

of us participating in the ORT, operations.

Compared

this

was our

first

to the challenges Gruel

for those

experience with nominal

had thrown

at us for the

past few months, operating a deep space probe twenty-four hours a day,

seven days a

week with no

disasters

The second rover cruise healthcheck

months

after the first healthcheck.

was

a piece of cake.

was scheduled

for June 17, exactly six

While we had been conducting ORTs

SOJOURNER

260

and RORTs, testing and training with the Marie Curie rover and learning to

work

together with the lander team, Sojourner had been asleep, con-

serving its batteries, during the long trip to Mars. This "late cruise" health-

check was also our chance to load Sojourner with the software updates Jack Morrison and

We

Tam Nguyen had produced since launch.

loaded two sequences onto the Pathfinder spacecraft. After con-

firming the loads, the lander team activated the

would wake up the rover and

giYt

it

first

sequence, which

the forty-four software patches

had approved. The waiting was more excruciating than the

first

we

health-

check. Back in December, the travel time for a radio signal from the spacecraft

was measured

away

that a

to

make

in the tens of seconds.

message traveling

the

trip.

seeing the result

But the

So the

at the

results did

was so

far

speed of light took over nine minutes

minimum wait between sending a command and come

Sojourner was up and running. The

in.

command showed that

ing every software patch. After pleted and the rover shut

would take

Pathfinder

was nearly twenty minutes.

telemetry from each rover

sent

Now

more than ten minutes,

down once more. The

effect the

Sojourner was acceptthe sequence com-

software changes

we had

next time the rover was powered on.

A quick analysis by the rover downlink team indicated no rover anomalies.

Sojourner appeared to be as healthy as

At the Rover Coordinator's request, the

when she started her voyage.

second sequence was activated.

This was a typical self-diagnostic sequence, nearly identical to the one

had used

six

ment was would

also

months

still

before,

which would confirm

operating properly.

Normal

that the

demonstrate that the software patching of the prior sequence requisite time delay, the

telemetry downlink arrived. Sojourner had done what

without complaint, and then shut Sojourner's safety

we

was now

heard from the rover,

down

in the

it

we had asked

of

it

as planned.

hands of the lander team. The next

would be on the

to suppress the nagging thought that again.

APXS instrucommands

execution of the

had not broken the onboard software. After the

time

we

we might

surface of Mars.

I

tried

never talk to Sojourner

TWENTY-THREE

MOMENTS

the last few

OF TRUTH

weeks before July 4, 1 found that

I

just wasn't hungry,

and

In began losing weight. We had reached the point where there was nothI

ing to do but worry over whether the landing was going to succeed. Afthere were a lot of things that could go wrong.

ter

all,

be.

No one really knew what the chances of a successful EDL was a complex process. For example, there were

landing would forty-one indi-

vidual pyrotechnic events (actually small explosions), every one of which

had to happen in its own time during EDL. Each pyro or separate a part of the spacecraft

toward the Martian surface; the

when

failure

firing

would release

appropriate during the descent

of any one of these would

doom

the landing.

cur

One of the key elements of EDL was "aerobraking." when the lander, protected by its heatshield, entered

mosphere percent of

at its

This would octhe Martian

about seventeen thousand miles an hour, and then

at-

lost 95

velocity in just a few minutes. Even the thin atmosphere of

Mars would provide

sufficient

drag on such a fast-moving blunt object to

decelerate the lander to a speed

where the parachute could be

safely de-

ployed.

Rob Manning was

the Pathfinder flight system engineer, the lead en-

SOJOURNER

262

gineer responsible for to operation. In a

fit

aspects of

all

EDL, from design

to implementation

of gallows humor, he considered an alternate braking

scheme. Just as an "atmosphere" was the gas layer that enveloped a planet, so a "lithosphere"

ning coined a

was the top

new term

ros failed to fire

on

that

would come

Of

it hit.

rain features discovered

sponded that Pathfinder

if

Man-

The

Mars would

surface of

no matter how

fast

it

was moving

Tim team member who had been wondering about how ter-

course, there

Parker, a science

surface.

into play in case any of the py-

time: lithobraking.

effectively bring the lander to a halt,

when

—the

of a planet

solid layer

its

little left

of the lander

.

.

.

on Mars by Pathfinder would be named,

lithobraking

would be

would be

was used, the only new

impact

ter lithobraking, the rover

crater.

The

feature

named by

rover team chimed in that,

would be transformed

re-

af-

into a subsurface ex-

plorer vehicle.

Lithobraking started out as a joke, but as landing day approached, our

nervousness

made

it

seem less funny

Outside the Mission Support Area there

showing the timeline of operations timeline indicated

when

critical

was

activities

a chart covering the wall

and associated

spacecraft activities

would

staffing.

occur,

The

when

commands would be sent, telemetry received, and when individual team members would be on station. The detailed plot of this information showed events happening twenty-four hours a day for several. sols. The timeline

was updated

for each Operations Readiness Test

and

titled ac-

"ORT 6," "ORT 7," etc. One day walked by and saw that the timeline was now labeled simply "FOR REAL." cordingly as

I

There would be no more

tests.

July 4,199?

I

was watching

area,

about

NASA TV on

fifty feet

a

monitor

in the rover engineering analysis

from the Mission Support Area.

NASA TV was run-

— Moments

ning a

VOCA voicenet twice: tor. If

MSA, focused mostly on Rob Manning. The

feed from the

live

was

over the

first,

also

turned on, so

VOCA,

we were

there were cheers or applause, we'd hear

was nervous

watch.

I

as hell,

from the

them three

and there wasn't anything

Rob Manning continued

to describe

the spacecraft. His narration told the world

pened ten minutes tectable,

earlier;

would only just be

was just

TV moni-

times, because

directly.

wondered whether I'd have a job to do tonight.

the landing.

It

hearing what was said

then, a few seconds later,

they were loud enough to be heard I

263

of Truth

for

me

It all

to

do but

depended on

what was going on with

what should already have hap-

the signal from the spacecraft,

if it

were de-

arriving as he spoke.

after 10 a.m., Pacific Daylight

Time. Sharon Laubach, the

who had trained to review rover command sequences, was sitting next to me as we watched the video feed. was sure that we both had the same sick expression I saw on her face. When we Caltech graduate student

I

were about ten minutes from possibly receiving the landing telemetry

—

which meant Pathfinder was hitting the surface now and

said,

"I'm sure everything

is

going to be okay."

I

turned to Laubach

"Is that a

premonition?" she asked. "Premonition." For some reason,

question or a I

had just

de-

cided the landing had worked.

And the next few minutes proved it. We had expected to lose the radio signal from the spacecraft at the moment the parachute deployed. Instead, we detected it almost all the way to the surface. A few minutes later, Sami Asmar, the communications engineer who had been flown out to the DSN tracking station in Spain, reported in. "Comm, this is Madrid. I ." Pathfinder had survived the landing! Cheers went see a weak signal up on the second floor of Building 230. We had thought we would have to .

.

wait hours for confirmation, but the signal came through almost immediately.

And

the length of the next signal told us that the spacecraft had

landed right side up,

"What

sitting

on

its

are the odds!" he bellowed, a

base petal.

huge grin

Rob Manning whooped. on his face. Of course, the

odds were about one in four, since Pathfinder was a four-sided tetrahedron. So

far,

Pathfinder had been lucky.

Four hours

later

it

was time

for the first telemetry session

from the

SOJOURNER

264

lander. If

had gone

all

our lander would have retracted the spent

well,

had cushioned its impact on Mars. Then the petals would have

airbags that

opened, revealing the lander's solar panels and exposing the rover

itself to

an alien sky and the soon-to-rise Martian sun.

DSN had locked onto the low-gain transmisEveryone in the MSA was ecstatic. "We have rover

"We've got lockup!" The

from the

sion

data!"

lander.

Matt Wallace announced over the voicenet.

The lander had switched on

the reed-relay waking

up the

Mars Local Solar Time (MLST). Healthcheck data came

6:59

Some

rover.

realized

we were

seeing

data conversion problem he had seen during the past few months. "The data.

first

The next one should be

okay."

is

a

commanded

bad.

We

The engineering

same

the

Curie over

can

t

trust the

analysis

team un-

know that the bad initial data was not

evidence of a hardware problem on Sojourner.

from

symptoms of

ORTs with Marie

healthcheck

derstood the system well enough to

rived

from the

of the data looked wrong. Suddenly one of the engineering

team members

analysis

in

rover, at

healthcheck

Then

at 7:35 a.m.

the telemetry

ar-

MLST. Sojourner

looked healthy!

Lander data was coming

in.

The

tilt

of the lander would be

deploying the ramps and getting the rover

off. If it

were too

critical to

steep,

we'd

have to adjust the petals to reduce the angle to something the rover could

One of my nightmares had always been that the landing would

negotiate.

go perfectly but the rover would never be able der was

.

.

two

.

degrees! Pathfinder

was

to drive.

The

tilt

of the lan-

sitting virtually level

on Mars.

We hadn't seen pictures yet, but how much more could go right?

The biggest uncertainty for the Pathfinder project had been whether the previously untried technology for Entry Descent, and Landing

work. With a successful landing, most of the Pathfinder team was elated

.

.

.

team was

and beginning to thrilled,

the landing barrier

elated,

relax.

thrilled,

Everything else was gravy. The rover

and nervous.

No

one had

would succeed; the uncertainty had stood

really

known

now down. The

if

as a psychological

between us and the mission we would perform with the

That barrier was

would

lander hadn't blown

its

rover.

big moment.

Moments

Now we

the rover team's big

wouldn't screw

Knowing I

it

moment was

that Sojourner

should try to get some

to

approaching, and

fast

had weathered the landing,

I

after 6 p.m.

borrowed the key

I

also

knew

that

My shift wasn't until that night, and the first

sleep.

to Art

want

didn't

I

mess up the most important job of

hausted.

we hoped

up.

images weren't due in until

want

265

of Truth

my

to leave, but

life

by being

I

didn't

totally ex-

Thompson's RV, which he'd parked

near the Space Flight Operations Center as a makeshift bunkhouse for rover

doze I

team members, and forced myself

few hours.

for a

I

off. Finally,

was

tried to sleep for I

some

headed back to the rover control

in time for the first high-gain

Gordon Wood,

enough

aligned well

we'd be stuck with

communications pass from Mars.

warned

might never get

that the high-gain antenna

to transmit a

good

signal to Earth. If

a low-data-rate mission.

in the best

way

possible.

to

area.

The

signal

he was

came

The IMP had found

right,

in strong,

Wood

stronger than even our most optimistic plan had predicted.

been wrong,

manage

the lead lander communications engineer, had

months before

the project

to leave the center of the action

time, but just couldn't

had

the sun, the

high-gain antenna was pointed straight at the Earth, and the

first

images

were coming down! Pictures from Mars! The rover team was overflowing the engineering analysis area, staring at the monitor in the corner. Cheers

came from

several spots

monitor with

around the

a cluster of people

through a peephole

at a

new

floor,

everywhere there was a

around it. The pictures were

world.

Some of

so

narrow

rocks, parts of the lander. that

it

was hard

to

difficult to interpret.

The view from

know what we were

wondered tion to

if it

were

a

and

felt

can drive

what

a rock

to!" Brian

scale that

its

would appear strange

behind

it. I

said

Wilcox wasn't so

rock was:

Was

it

it

IMP camera was at.

place.

I

Before you

saw a piece

to

And I

I'd

I

never paid atten-

wasn't a surprise that

something

sure.

We

a surge of anxiety while

damaged or bent component.

the lander

view of sand with

idea

didn't recognize,

raw IMP images during any of the ORTs;

some images of

we

I

the

looking

could figure one image out, the next one would take of the lander that

looking

like

the pictures were highly

compressed, so they were particularly blocky and

saw sand,

TV

me. There flashed like

a

"There's a place

realized that

I

had no

big and far away or small and close?

SOJOURNER

266

When

There was nothing recognizable to provide context. engineers had assembled

all

of these "postage stamps" into a mosaicked

panorama, then we'd know what we were

Some of

seeing.

those images went together to form the airbag assessment

panorama. Just

as in

rover petal. Here ticed solving.

one of the ORTs, there was airbag material on the

was

a

problem the Pathfinder operations team had prac-

Someone on

lander sequence to

lift

team pulled up and modified the old

the

the rover petal

up about

down

sent.

We

would

again.

ORT

run the

forty degrees,

airbag retraction motor, and lower the petal back

mands were

the imaging

The com-

see during the next communications pass

whether they had done the job.

We

any ideas

was rushing

why we had

the rover sequence.

dow to

assumed

I

Pathfinder had closed

before the

DSN

it

that the

down before

I

difficulty

it

ex-

had

commands

in

came down, or

had been sent down

didn't see

how

to find out

downlink opportunity pass

showed us

that

something was

at least

defi-

not well. As the existence of the communications

spread through the team,

all

of our fears began to surface. The

telecomm subsystem was the area most of us had had the vations about.

The only thing we knew

we would not be

for sure at the

deploying the rover on sol

The Pathfinder

folks

were concerned

success after another, almost as

if

greatest reser-

moment was

that

1

too.

So

someone had

far,

sol

1

had been one

expertly choreographed

Even the draped airbag problem had been neatly solved by the recovery

sequence, as proven by the latest

problem was the

first

IMP images. The

rover communications

sour note. Jake Matijevic was forced to go off to the

Pathfinder press conference at the worst possible time: a problem; fix

I

communications win-

the other data

signal.

if

wrong. For some reason, the rover and lander were not talking to

each other, or

it.

we had

somewhere and wondered

was just

had locked onto the

until the next

data from the rover as

had been missed because

The next communications nitely

off

only gotten data from a few of the

that the rover telemetry

more

much

hadn't gotten quite as

pected. Matt Wallace

it,

it

could

We knew we had

mean the end of the Sojourner mission if we

couldn't

and the rover team hadn't even had time to hypothesize on the

Moments

With his chronic asthma,

cause.

tonight as he headed fled,

26?

of Truth

Matijevic

had never been high-energy, but

down to von Karman Auditorium, he

stooped over and feeling the weight of the complete

actually shuf-

failure that his

system engineer's mind could not help but anticipate.

The

now

Lin van Nieuwstadt, had flown

operations. Since she ing, the

work

rover team stayed behind to

in

the problem. Lin Sukamto,

from the Netherlands

had been absent during the past

six

to support

months of train-

most she could do was go around and ask questions about what

was going on. Now, with the radio modems she had once been ble for in question, she

providing answers.

Even Donna

was suddenly

We

Shirley

all

had joined

Everybody was offering

in a position

crowded

their

us.

own

responsi-

where she needed to be

into the engineering analysis area.

The downlink team gave theories as to

their report.

what was causing the

problem. Van Nieuwstadt thought that the oscillator crystal had

Someone

tured on landing. silent state

flowed,

I

we used when

my

added

two

else

wondered

doing cents:

APXS

if

frac-

the rover had gone into the

data collections. As the theories

"Everybody

is

tossing out ideas about

what's wrong. Let's start writing them down, categorizing them, and eval-

The problem could be caused by hardware,

uating them.

even an operations procedure. Let's get

and determine in

spending a

we

can do to

all

how we're going to investigate of time on the hardware

lot fix

them.

And it may be

crazy,

there

Is

some way that the

data, or that

it's

down

each one. There's no point

failures, since there's

nothing

way

that the

but

is

there any

problem could have nothing to do with the rover? Could der?

software, or

the possibilities written

it

be on the

lan-

lander could be incorrectly processing our

got the rover packets, but

it

just isn't sending

them

down?"

The ing on

than

I

late-night pizza arrived.

on Mars, had

in

I

As

I

thought through what might be go-

my appetite had returned. felt hungrier grabbed my share of pizza. It occurred to me that

realized that

weeks.

I

I

I

should be depressed about the impending everything stead, solve,

I

I

had devoted so much time

was happy. The waiting was

doom

of Sojourner, the end of

to for the previous four years. In-

over.

Here was

a real

problem to

not the hundreds of imaginary ones that had been plaguing

me

for

SOJOURNER

268

months. Solving problems, or helping others to solve them, was what

was trained

for.

This

I

knew how to

deal with.

Jan Tarsala wasn't assigned to work his

He came

The

first shift

pizza

was good.

on the rover telecommu-

JPL anyway

nications

team

tion was.

He stayed around to see the first images arrive from Mars. home to his family and had dinner.

late

until sol 2.

into

I

where the

to be

ac-

In the

afternoon he went

About went out

eight-thirty in the evening the telephone

to the kitchen to

Nieuwstadt or Scot

answer

No

Stride.

it.

one

would be

else It

radio.

It's

He

either Lin

van

him on

the

calling

was Lin on the other end, "and

she was crying. She was extremely distraught."

"The

house.

at his

He knew it would be

He picked up the phone.

Fourth of July.

rang

Van Nieuwstadt

said,

not working."

"What do you mean it's not working?" "We're hardly getting any packet throughput. transmit and

We

transmit and

we

we transmit, and nothing s coming across the data link." Van herself. "What are we going to do?"

Nieuwstadt was just beside Tarsala before.

remembered well

"Hold on. Hold on.

not a big deal.

It's

the predictions of Jim Parkyn, nearly a year

It's just

off-frequency

We can manage that. It's

going to be okay." Van Nieuwstadt feared a catastrophic

of the radio. Tarsala tried to reassure her that

failure

ting through, then the radio itself

push those radios.

crystals

We've got

around

a heater

system available to

us."

was

if

something was

get-

we

can

fine. "If it's off-frequency,

frequency by heating, or not heating, the

in

on each end of the

link.

We have that control

He was convinced that they had the means to con-

trol the crystal frequencies to get the lander and rover matched

talking again.

"We can make

this

Van Nieuwstadt had been

work."

totally

uninvolved in the rover since her de-

parture from JPL about nine months before. Yet she had so self personally invested in the

the worst of situations: She tions,

much

of her-

Sojourner radios. Van Nieuwstadt was in

felt fully

but she was facing the

up and

responsible for rover

crisis cold,

preparation over the prior several months.

communica-

without the benefit of careful

.

Moments

The rover team gathered for

of Truth

269

midnight brainstorming session. Donna

its

down

Shirley offered to be the recording secretary, writing

blackboard. After an hour of discussion, the board

and annotations

scenarios, sol

1

filled

the spaces

was

between the

with possible

lines. Initially

on

(and before that in cruise) the rover telecommunications subsystem

had been performing just

fine.

Then communications degraded

most nothing was getting through, except frames.

We needed to be

Just because the link, this

careful not to

symptom of the problem was no

itself.

Early in the meeting,

assumptions:

coming

across the

was the telecommunications

from communicating properly.

all

of the old problems had resurfaced as possi-

These were the ones that the rover team had

the design, development, and test phases and had in ized, corrected, or built

a

al-

Perhaps something had gone wrong elsewhere on the

rover, preventing the rover

ble causes.

data

until

communications

for garbled

make unwarranted

did not guarantee that the culprit

subsystem

done

filled

on the

ideas

identified during

most

work-arounds to deal with.

cases character-

What

good enough job? Someone mentioned the rover

we

hadn't

CPU

crystal

if

problem, which had once caused the Marie Curie rover to run in "slow motion." Henry Stone insisted that

this

was highly unlikely. The

electron-

And besides, even when the problem had existed, it only occurred when the rover started up. "The rover never started out okay and then went bad!" And that was just what ics

team had redesigned the clock

had appeared to happen during

Could

sol

circuit.

1

a temperature difference

between the radio modems

in the

lander and the rover have caused the two radios to be transmitting at ferent frequencies? This

compensated installed.

crystal oscillators

would have

that the temperature-

solved, if they

The telecommunications team had examined

seum over likely

was the same problem

the prior

two

years. Lin

cause of the communications

van Nieuwstadt failure.

dif-

had been

the issue ad nau-

didn't see this as a

Jan Tarsala had not been told

of the brainstorming session, and the rest of us had no inkling of Tarsala's confidence that temperature-induced frequency

drift in

the radios

was

in-

SOJOURNER

2?0

deed the root of the problem. The impression in the room that night was that while this

might have contributed to poor communications,

seem

account for the almost total loss of data transfer

likely to

it

didn't

we were

seeing.

The

APXS

rover didn

t

have enough power to operate both

radio and the

its

we had designed a "silent mode" that turned off communications during APXS data collection. Most of us had been a bit skittish about the rover's silent mode ever since we'd come up with it. Any time you sent a command to tell the rover to stop listening for new commands, you had to be careful. I knew that every sequence we sent to the rover was supposed to turn the silent mode off before it completed, at the

same

time, so

whether or not we'd ever turned communications-restoring

was somehow

the rover

nications frames

it

on

in the first place.

data received by the lander

would have

You might expect

lander.

modem power

likely,

ply

was one of the few

since there

so any

coming

few random

a

buy

this theory.

failure?

Could the

didn't

component

supply have gone bad? That didn't seem

were a few frames getting through. The power sup-

too

since there

all,

that the lander's diagnostic

The communications team

there had been a hardware

lander-mounted

talking at

to be spurious radio noise

bad frames, but nowhere near the number

if

the

command had somehow been skipped? And if mode, where were the garbled commu-

from some other source on the

What

if

in silent

coming from? The rover wouldn't be

software had recorded.

But what

areas

were two power

one together with the

where there was

supplies,

modem heater.

a

redundant component,

one to operate the If

we

modem

alone,

ran out of other options,

could always switch over to the unused power supply and see

if

we

the prob-

lem went away.

The

severity of the

What had changed on was

problem had grown worse

the lander during that period?

that the lander's high-gain antenna

the Earth

all

tions,

began

progressed.

The obvious

thing

had deployed and was tracking

as the high-gain

antenna moved to

to scatter or block radio signals

mounted rover antenna and as

1

day long, maintaining communications between Pathfinder

and the DSN. Perhaps it

as sol

Sojourner, getting

time went by. Since the rover was

still

new

orienta-

between the lander-

more and more in the way

stowed on the lander petal,

its

an-

Moments

tenna was also

down

lying

still

of Truth

271

along the solar

mounted antenna was deployed and

array,

while the lander-

two antennas

vertical, leaving the

perpendicular to each other. Although data had gone back and forth be-

tween the rover and lander during ground orientation

relative

sible

weakened the

link,

then

of the antennas.

the rover unstowed and

was the worst pos-

other conditions had

If

might have pushed the telecommunications

this

subsystem over the edge.

testing, this

also

It

meant

we managed

that once

to get

antenna upright, the current problem might

its

go away. Other than one or two suggestions and questions, there was

for

little

me to do at the meeting but listen and look for holes in the logic. The subsystem experts were these symptoms.

I

sitting in that

stories

I

inventing hypotheses to explain

at

had done my job: nudging people

then standing back to

be

than

far better

let

them do what they

meeting. For most of

of JPL engineers

who

in the right direction,

did best. Yet

my life,

I

was proud

to

had been amazed by the

I

could magically figure out what was

wrong

with spacecraft far out in space and, even more impressive, could fix them!

Here

I

was, and

was making

I

group of engineers. munications.

people on

I

this

didn't

I

didn't

know

know

if it

discussion.

trude periodically, announcing all

if

small contribution to just such a

we would be

could be

team were the only ones

were focused on the

With

my own

of the options

But

fixed.

to

able to fix rover

do

it.

if it

was

Most of

possible, the

my

my mind love my job!"

But a small part of

silently,

listed,

"Damn.

I

the next steps

would be

thoughts

would

still

lier.

Lin van Nieuwstadt could barely keep her eyes open. She was

2 a.m. Matijevic ordered himself

team would think better

after

something

like a

After the exhausting day and night of sol

home most

news

to report

on

5

and turned on the

TV

tonight,

good 1,

of the next day and get some

morning on July

ear-

and the rover team to go

home. There wasn't much more they could do

stay

rest as

on Netherlands time, so she should have been in bed twelve hours

Around

in-

to investigate

each one and determine which could be discounted, leaving the possibilities.

com-

and he

felt

the

night's sleep.

Matijevic had rest.

hoped

He woke up

to

mid-

Since there was no other bad

Pathfinder, the only topic

seemed

to be the silent rover.

There were speculations that the commercial radios were no good. Mati-

SOJOURNER

272

and shook

jevic sighed

rover

He'd better get into work and shield the

his head.

team from the media so they'd

problem

.

.

When

the

telemetry in

pened

actually get the time to solve this

.

first

downlink

A total of 31,491

it.

came

for sol 2

there

in,

was

lots

of rover

bytes had been transmitted, which hap-

to be the capacity of Sojourner's telemetry buffer. Apparently, dur-

ing the period of failed communications to execute

its

command

telemetry until

it

on sol

1

the rover had continued

,

sequence as ordered, dutifully storing away

its

overflowed the buffer memory. We'd lost whatever data

the rover had generated after the overflow, but at least

we were

talking

again.

When the press heard that the rover was working, they wanted an explanation.

The

real

answer was that

we

didn't yet

know. But between the

media's need for an answer and the attempts to accommodate by very

pressured engineers, pure speculations were transmuted into apparent facts.

The

story

went

"hitting the reset

that the rover

had been

fixed

by the equivalent of

button on your computer a few times." This version of

the truth spread so far and so fast that even spokespeople at JPL, as well as

members of the Pathfinder team, believed it and were repeating it to the press. I knew that we hadn't yet had the opportunity to do anything unusual to rectify the problem. The only "resetting" of the rover was due to the normal

command

have the rover shut

sequence already onboard, which was designed to

itself

temperature readings.

down overnight, waking up once per hour to get

doubted that that had fixed anything.

I

Scot Stride winced

when he heard

telecomm subsystem was

his,

Nieuwstadt's, and as far as he

the press reports.

Jan Tarsala's,

was concerned, they were

of assessing the problem. Stride's

own

The

still

in the process

guesses as to the cause

from the knowledge that they had never had the chance

stemmed

to test rover-

lander communications in the particular lander configuration that existed

on Mars: The rover and

its

test,

but

this

at that

now

antenna were stowed with the lander

petals open, sitting in the middle of a

had come to

rover

Sami Asmar's, and Lin van

wide open

terrain.

The

closest they

on Earth had been during the system thermal /vacuum time the lander was enclosed by metal chamber walls,

which caused the radio

signal to

bounce

all

over the place.

Moments

of Truth

2?3

Whatever the cause of the communications reason that communications had been regained,

loss, it

and whatever the

was now time

to de-

ploy the rover.

Justin Maki

was

that put

cially,

a research associate at the University of Arizona. Offi-

him on

the faculty, but in reality he

from graduate student slave

labor. Peter

was barely one

step

Smith had hired him to build IMP

camera imaging sequences during the mission, and brought him out

JPL with the

of the

rest

IMP team

twenty-seven, just out of school,

He

sion!

couldn't believe his luck.

to

up

do the

job.

to

Here he was, barely

and he was part of the Pathfinder mis-

He had a VOCA unit on his desk,

so he

could hear everything that was going on in the Mission Support Area, and his office

was next door to Rover Control. Pathfinder was on Mars, the IMP

camera was working like to do.

It

Well, there

Maki

a

champ, and

his

sequences were telling

was one thing he did want

figured that

it

what

much better than this.

didn't get

if

he knew

how fast the

to

do

—make rover movies.

rover moved, and he could get

the timing close to right, he could track the rover with the IMP, and take pictures a

few times

a

minute while the rover was

tos together, you'd get a kind of jerky

driving. If

you put pho-

movie showing what the rover

did.

The

resolution of the images wouldn't have to be so good, either, so the

data

volume

ations

hit

team might

value. But

tific

would be

small.

The problem was,

the Experiment Oper-

no

see such rover movies as superfluous, having

Maki knew

that rover movies

would be

a great

scien-

way

document what Sojourner was doing on Mars, and would probably ture the imagination of the public.

fun to see

if it

when

it

would be

the last few Operations Readiness Tests were

going on, Maki had approached

sion

that,

cap-

could be done.

Before landing,

Planner. Dias

More important than

to

was responsible

would unfold

Bill

Dias, the project's primary Surface

for designing the scenario of

after landing.

how

the mis-

He was also responsible for budgeting the

downlink data volume between the competing sources of telemetry,

which included the lander engineering subsystems, the IMP camera, the lander weather experiment, and the rover. Dias liked the idea of rover

SOJOURNER

2?4

movies. In

volume

he had always maintained an extra margin of telemetry

fact,

in case

more imaging of

Dias and Maki schemed.

the rover proved necessary.

What was

the best

way to

movies into the plan without being too obvious about the idea quashed?

of

its

incorporate rover

which might get

it,

A rover movie would have to be a command sequence

own. Every sequence needed

a

unique identifying number. For ex-

ample, the imaging of the rover taken every day to update the rover's position

was always

reviewed

at the

Anything too

in

sequence S-0055. The status of

sequences was

all

Mission Plan Approval meetings scheduled for every

different

from the other sequence numbers might stand out,

Maki and Dias went through the

calling potentially fatal attention.

quence assignment

sol.

list.

The 50

se-

was mostly rover-related sequences.

series

S-0053 and S-0054 were reserved for soil mechanics experiment imaging, S-0055

was

taken, and S-0056

leave in the Martian

was

for

imaging the tracks the rover would

S-0050 was available.

soil.

Maki

'Rover Navigation Imaging/" That sounded a lot

smiled. "Let's call

more

it

respectable than

"Rover Movie."

During ORTs 6 and rover egress

down

7,

the ramps. Getting the timing right

pretty easy, but aiming the tried

it,

Maki had experimented with movies of the

IMP was

a bit of a problem.

he caught only the top half of the

aiming the engineering model of the

rover,

seemed

The

first

to be

time he

then only the wheels.

And

IMP camera was a bit different from He would just have to give it

aiming the camera on the actual spacecraft. his best shot

We had

when the

real

opportunity came along.

commanded

uploaded the sequence that

was looking

at the

monitor

telemetry started coming

up on the

screen,

in.

the rover to stand up.

in the engineering analysis area

The IMP image of

the rocker latch flashed

I

couldn't

vouch

for the right rocker,

which could not be seen with the lander camera, but the fine.

I

the

and was gone, replaced by the next picture. But the latch

spring had been straight and true.

ployed just

when

My

worst nightmare looked

like

it

left side

had

de-

would remain only

that.

But then the engineering telemetry from the rover

itself

indicated an

Moments

Henry Stone was

error.

remember this was his job

as

of Truth

275

the Data Controller

on watch, and he would later

one of his personal denning moments

to decide

whether the rover stand-up had been

Thompson

him over

your

call,

were

to get the rover off the lander during sol 2,

Henry," Art

cision soon.

for the mission.

said to

VOCA net.

the

if

sure!

He was

its

wheel

with an un-

restraints

latched rocker, there

was no going back: There would be no way

that rocker in place.

The

see,

was being reported

compared

what we had expected. But

to

was the

were so cautious

in

first

our

it

time the

that

there.

that the left

became

clear that

we

could not

for the side of the rover that

and indicated that the encoder on the bogie had

fact that this

knew

of the rocker-bogie, Henry

rocker had latched properly. Digging into the data, the error

to lock

would be over then and

rover mission

IMP photograph

the

we

he made the wrong choice, he could be the goat of

the mission. If the rover backed out of

From

If

we'd have to make the de-

But Stone didn't have enough data yet to be

keenly aware that

It

successful. "It's

shifted a

few counts

might be explained by the

FUR had stood up in Mars gravity. We

test for success that

only a slight variation in the

expected readings would cause the rover to stop what

it

was doing and

human intervention. "Okay, Henry. Don't be rash," he told himHoward Eisen was already convinced that the rover had unstowed

wait for self.

and was vocal about that opinion. While Stone was

successfully,

happy with agreed test

Eisen's certainty in the perfection of the unstow,

that, if the

still

not

he generally

unseen rocker-bogie had not locked, the backup latch

would have caused

skewed appearance

that side to sag noticeably, giving the rover a

in the

potentiometer readings.

images and unmistakable discrepancies in the

None of

that

had happened. Stone declared the

rover healthy and ready for egress to the surface. Earlier,

ramps

Brian Cooper had been studying the stereo

in the

RCW display,

ward lander ramp was

and had come to the realization that the

sticking out into the Martian

diving board over the ground.

down

the other ramp.

choice.

It

led to a clear,

We were

open

air,

suspended

rear

ramp looked

like a

its

like a

good

area almost completely devoid of rocks.

The one rock in the vicinity, which had been immediately christened nacle Bill" for

for-

going to have to drive the rover

Going down the flat,

IMP images of the

mottled appearance, would be

a perfect first

"Bar-

rock target

SOJOURNER

2?6

for the rover's

Justin

APXS. Rick Welch built the modified

Maki designed the

first real

tended to capture a special

"rover movie" sequence for the IMP, in-

moment on

Mars, with the timing set to cor-

respond to the delays inherent in the rover

no

sophisticated software tools to help

command sequence. Maki had

him

select the appropriate point-

ing parameters for the IMP, which depended on

was

tilted. Instead,

in the existing

the

IMP

he eyeballed

ramp

to point

it

it,

how steep

the rear

ramp

counting the number of rows of pixels

pictures to guess

right at the

rear-egress sequence.

how much further he needed to

tilt

end of the ramp. The sequences were up-

linked to the lander.

When went

tion

the thumbs-up

mands went up team waited

The

came from

to the Coordinator to

Flight Director.

queue the sequences already onboard the

The comThe

lander.

for the confirming telemetry.

rover movie images started to

the other IMP images. Every one

waited

the downlink team, the authoriza-

and on to the

intently.

Had he timed

come down, mixed

in the

MSA watched

the imaging correctly?

together with

the screen.

Maki

Had he aimed

the

IMP at the right spot? The first image told him that he had. The end of the ramp was in the center of the view. The rover would have to come this way to reach the surface. But there was no rover in the picture. That wasn't too surprising: Maki had designed the sequence to have two images before and after the rover should have driven through the scene, just in case the timing

peared.

Then

Next one.

Still

was

the next

no

off.

A

few IMP photographs of other

ramp image

rover.

flashed up,

Maki worried

again.

still

with no

targets ap-

rover. Okay.

Had he missed it? The

rover

was late. He did not want to have messed up. Another member of the IMP team from the University of Arizona pointed

a finger at Maki.

screwed up. You screwed up!" he repeated, an odd smile on

Meanwhile, Brian Cooper looked Shit!

out:

Where The

at the

same

pictures

"You

his face.

and panicked.

was the rover? In his mind, the worst case scenario played

rover had rolled off the side of the ramp, flipping over to

manently upside down

in the

Martian dust.

Where was

At the same moment, Henry Stone stared intently screen in the analysis area.

He was

lie

per-

it?

at his

workstation

waiting for the rover engineering

telemetry that would either dispel or confirm his lingering doubts over

Moments

the success of Sojourner's stand-up. latched?

Then

they'd be screwed.

"MOVE" command,

2??

of Truth

What if the right rocker really hadn't Some telemetry came in. The first

driving Sojourner several inches onto the rear ramp,

had executed normally. There was no error rocker-bogies must be fully latched, or the

He had made

report. Stone relaxed:

command would

have

Both

failed.

the correct choice.

Art Thompson, the Rover Coordinator on duty in the

watching the engineering telemetry stream, waiting for

it

MSA, was

also

to confirm that

commands had executed without error. The Flight Director was impatient. "Rover. Can you confirm?" The rover telemetry stream had stopped at command 2679. That only told Thompson that the rover had backed halfway down the ramp without detecting a problem. There was no more telemetry! The damned communication problem was still there. Thompson pondered what he was going to say. He really, really didn't want to have to say he didn't know what the rover had done. He paused. Someone told him to look up at the monitor displaying the IMP all

of the egress

images

as they

There

it

came

was!

in.

The

rear left

wheel appeared

bottom. Then the wheels were on the ground.

rover

was nearly

nally,

an image flashed up with Sojourner

The

the surface.

at the

MSA was

filled

into his headset microphone.

The

The

in the next image.

fully off the

with cheers. All

"We can

ramp and

right!

safely

Thompson

late:

As

on

spoke

report visually six wheels on

rover had indeed been about a minute

Fi-

soil!"

a consequence of

the rover-lander communications problem the rover had spent

more time

down

the ramp.

than usual attempting to communicate before

But

now

first

thing

it

sat firmly

it

did

on the ground, ready

it

drove

to carry out

was plop the APXS down onto the

soil

its

mission.

and begin

The

collect-

ing data.

Maki pieced together the ramp images movie, and

made

it

available

that comprised the rover

on the Pathfinder network. He went next

door to Rover Control and told us to take a look rover movie and played

it

on one of the

down

it.

RCW displays.

frames, repeated over and over again, but

rover drive

at

we

We It

called

up the

was just

didn't tire of

a

few

watching the

the ramp onto the surface of Mars. The movie revealed

one thing that nobody caught when the images were displayed one by one

SOJOURNER

2?8

during the downlink: In the frames before the rover came into view, you

could see the ramp flex under the weight of the advancing rover. As

someone from

we were public,

the

we

had,

MSA rushed in and told us to get out of the way,

that

all

could from the few images that

blocking the view JPL was sending the rover movie out to the

and the way

it

was being done was

this:

video feeds in the Pathfinder Mission Control area in the big conference

room, and one

There were three

—one

in the

operated pan /tilt platform, "HAL9000" because

its

big

reminiscent of the computer's "eye" in the movie 2001

The camera

MSA, one

operator, wherever he was,

:

a remotely

zoom lens was A Space Odyssey.

had noticed the rover movie on

RCW display, and had zoomed in for a close look. Of the three avail-

able feeds, this

to

live

Rover Control. The rover team had

in

dubbed its video camera, mounted in the back of the room on

the

we

we

were attempting to glean

NASA

was currently the most

interesting, so

it

was being fed out

TV, and from there to the public at large. So, unbeknownst to

those of us in Rover Control, the public was looking over our shoulders at the

same thing we were. Or

they were trying

at least

to.

Apparently,

Sharon Laubach was standing in front of the SGI, partly obscuring

HAL9000's view of the movie. And,

thus,

someone was dispatched

to

keep us from interfering with the show!

Laubach found out

later that

her brother had been watching.

in the U.S. Navy, stationed in Japan.

NASA

feed,

woman

would

sister?"

ask, "Is that

her." Finally,

He and his buddies were watching the

and he had told them that

Pathfinder. Every time a

your

somebody

all

his sister

appeared on the

TV

was working on screen,

and he would respond "No,

said, "Is that

of the rover?" "That's her

He was

your

At

right!"

sister's

least

someone that's

not

nose blocking our view

her nose was famous, seen

around the world.

At the sol 3 uplink

planning meeting, Jan Tarsala explained the radio

frequency-drift hypothesis, and again, off-again

lution

would be

how

it

could fully account for the on-

communications performance we had observed. The

so-

to properly control the temperature of the rover's radio.

The big question was whether

to

run the heater to

warm up

the

Moments

rover's radio, or let the radio stay cool

didn't yet have

by leaving the heater

enough telemetry from the rover to be

him

right choice. His gut feeling told

recommendation the

2?9

of Truth

to

Henry

sure

He made his

Thompson, and me: "Don't heat

modem on the morning of sol 3." held up a sheaf of papers. "Here's the sequence.

I

up there

When

right now,

running on the

the rover wakes

I

by commands near the end of the

the rover ever asked the lander for

it

It's

will heat the radio for

its

way to comply

modem early on sol 3 was con-

sol 2

command sequence, before

new sequence.

had to be that way,

It

because the rover needed to communicate to receive the first

already built.

could do. There was no

with Tarsala's request. The heating of the trolled

It's

rover. I've already got the heaters on.

up tomorrow morning,

ten minutes." There was nothing

the

Tarsala

which was the

to keep the radio cool.

Stone, Art

off.

sol 3

sequence in

place.

So turning on the heater on the morning of

sol 3

would be

a kind of

experiment. Tarsala's bet was that the communications link would get

worse

modem

as the rover's

wrong," he said later. Sol

would show,

sols

3

got warmer: "Fortunately,

I

was dead

communications looked good. As the next few

the radio needed heating, not cooling.

"Once the radio

much better very quickly." would contain commands to pre-

exceeded a certain temperature, things got

Within a few

rover sequences

sols,

heat the radio before transmissions,

critical

transmissions.

Whenever

possible,

image

which would have long transmit periods, were moved

later in the day,

when

the radio

would

to

com-

naturally be warmer. Rover

munications turned manageable.

The post-sol 2 press conference was triumphant.

had been invited to be present

in

The

team

entire rover

von Karman Auditorium.

On

Earth, at

Warm though it was, was not going to deny myself the opportunity to wear my leather rover team

JPL,

it

jacket.

for a

rover eral

was

late in the

evening of July

Today we had earned the

5,

I

right to pat ourselves

few moments. As we gathered

on the back,

in the auditorium,

team was shaking hands. Donna

of us. She

1997.

Shirley

if

everybody on the

came over and hugged

made me turn around to show off the

only

unofficial rover

sev-

team

SOJOURNER

280

patch on the back.

I

on

he needed a prop during

stage, in case

Laubach

arrived with an 8X10 of the

cameras;

it

had just come

to Matijevic

down

gave the jacket over to Jake Matijevic as he sat

first

his presentation.

picture taken by

off a photo-printer near the

and then joined the

rest

of us

who had

by the Pathfinder lander mock-up on the west

Sharon

one of the rover

MSA. She handed it

already congregated

room. This was

side of the

the rover's night. I

had sometimes been concerned that Matijevic was not the most dy-

namic of speakers. He tended clauses. His condition

Tonight,

I

to speak in

complex sentences

full

caused him to cough often, and sapped his energy.

need not have worried. Matijevic was more excited than

He

ever observed.

of

spoke of the

new

I

had

planet to be explored and the fully

He surprised me further by actually holding up the jacket and speaking about the people who

functional rover that

was going to do

had made the rover

a success,

that exploring.

and then asked the audience to acknowl-

edge the rover team. The rover engineers stood to one side of the audi-

torium interspersed Matijevic's

among many members

commentary had

of the Pathfinder team.

shifted the attention to us:

As the audience

applauded, the video cameras turned in our direction, scanning the faces

of the team.

Layman during all of this? By the time of the Pathfinder Layman had been off working on another project for nearly a

Where was landing, year.

Bill

He had

not been involved with operations training, but had been

providing his special expertise elsewhere, where July

4,

he and

his wife

home. He had

told

felt

On

TV at

Eisen and others that he would be sitting by

way he

could,

if

there

was

a problem.

that he hadn't paid his dues during that year of preparation

for surface operations. "I

the

was most needed.

watched the landing and what followed on

Howard

the phone, ready to help out in any

Layman

it

woodwork

am struck by how many politicians come out of

to touch the heroes.

It's

like the

everybody wants to touch the quarterback.

would be fun

to bask in the glory.

But

trained throughout that final year

it

I

was

game-winning pass

had that same really

who ought

to

feeling, that

only the guys

be

there.

it

who had

Other people,

Moments

particularly

281

me, shouldn't be standing around,

He was

up."

of Truth

in the way,

mucking things

aware that for the bulk of the Sojourner design

also keenly

phase his decisions had been the definitive word on the rover, and had de-

termined the direction in which the team marched.

were on the

floor, in the

middle of things, people might turn to him for an-

swers and accept his responses as gospel, year out of date. the well-trained

Once more, the

He

He was afraid that if he

felt

when in fact his knowledge was

now less

he was

a

educated on the key issues than

members of the rover operations team. He stayed away. Bill Layman had made a decision with only the good of

team and the mission

in

mind, demonstrating the strength of

his lead-

way most would never recognize. though Layman would not allow himself to be present in the cen-

ership in a Yet,

ter

of mission operations, the success of Sojourner was, by his

oning, the high point of his technical career.

He had been

own reck-

the project

mechanical engineer on both Voyager and Galileo, JPL's flagship missions.

But those spacecraft, flying past and orbiting distant planets, just could not match the sheer emotion triggered by Pathfinder and Sojourner on

Time on the Pathfinder project was compressed, the events followso closely upon each other: landing, the first signal from the surface,

Mars. ing the

first

images of Ares

the rover driving

Vallis,

Pathfinder delivered huge, instant gratification.

done

it!

The team he had prodded,

been forced to leave behind

would this.

as

always

felt a

alien terrain.

The microrover team had

challenged, encouraged, and finally

he went to help develop the lander that

deliver Sojourner safely to the surface; they

Layman had

on the

had together created

great responsibility to the team, and had

never been sure that with the conflicting demands of the rover and lander

on

time that he had totally

his

fulfilled that responsibility.

complishment and release of years of

were

tears in his eyes.

rolling

"What

What really

Even

We

a thrill to

did

it.

first

thrill

of

ac-

were overwhelming. There

a year later, recalling the

onto the surface for the a rush.

effort

The

images of the rover

time, the tears nearly

came

back.

And nothing can ever take that away from

us.

have been the Chief Engineer on the rover! That was

something!"

TWENTY-FOUR

LIVING ON MARS TIME

Sol

3

"Something's screwy!" Henry Stone and Matt Wallace were not happy. I

had

just

walked into the engineering

on the

what

We were only just beginning to get communications link mystery. Now something else was

the rover's telemetry a handle

analysis area to find out

showed for sol 3.

know what had caused the probwe had commanded Sojourner to collect an

obviously wrong, and they did not yet lem. For the night of sol

APXS spectrum to

of the

2,

soil

behind the rover while

wake up the rover every few hours,

it

slept.

The plan was

store the intermediate

APXS results

memory for later transmission to the lander, do a healthdown again. This would satisfy the APXS scientists that their first APXS spectrum of an actual piece of Mars would not be ruined

in the rover's

check, and shut

by some

would

"glitch" that introduced noise into the data.

also give the rover

team

a set of temperature

spaced out over the night; from those data points, fine

The healthchecks measurements,

we would be

able to re-

our thermal model, and come up with a heating strategy to make

sure the

components

in the

WEB stayed within a survivable temperature

range.

But almost none of the planned night operations had occurred. In the

Living on Mars Time

late afternoon,

Sojourner had shut down, then collected

mode, and powered on

three hours in night

posed to shut down again for

When

of the night.

scheduled times for

journer skipped

p.m.

woke

APXS rover

data for

was

sup-

then power on to take an-

MLST, Sojourner went

the rover finally

to sleep for the

morning

to the

light,

the

of the nighttime shutdowns had passed, so So-

all

of those shutdowns. The temperature readings,

all

tended to be spread across the night, were early

The

again.

a couple of hours,

other reading. Instead, at 7:08 rest

283

all

bunched together

in-

in the

morning.

Stone was speculating that maybe there was a hardware problem with the rover's clock.

shut

down

explain the

He just didn't see

to an apparently

symptoms

to

an operational reason for the rover to

random

Once

time.

me, an obvious

I

had gotten them

possibility

came

to mind.

"What time was it when the rover shut itself down? Was it after the internally set

rover's

'END-OF-DAY' time? I think it may just have done an

shutdown." From the expression on Matt Wallace's

face,

to

auto-

knew

I

the

thought had never occurred to him. Meanwhile, one of the downlink engineers had gotten Jack Morrison

shutdown"

as the culprit

on the phone. He zeroed

even faster than

effectively inserted a "sleep until 8 a.m."

shutdown commands

I

in

on

"auto-

The auto-shutdown had

had.

command

in front of the other

in the sequence.

Now the question that remained was why had we never seen this beWhy hadn't auto-shutdowns interrupted our sequences during ORTs? We had done many overnight wakeups during testing. Morrison had the answer for this too. When the conditions were right for autofore?

shutdown (when time),

it

would

power was low and

solar

still

take about

more than enough time form

a healthcheck,

for the rover to

and shut

problem on the night of

two minutes

sol 2

itself

it

was past Sojourner's bed-

wake

up, readout

telemetry data in the

first

thing

it

its

did

still

buffer.

APXS

data, per-

down normally by command. The

could be traced straight back to the low per-

formance of the rover-lander communications

With communications

That was normally

to trigger.

spotty, the rover

had

link

we had been

built

up

a big

seeing.

backlog of

When Sojourner woke from its afternoon nap,

was send that data back to the

lander. This used

two minutes, and auto-shutdown took over before the

up the

commanded

shut-

SOJOURNER

284

down could execute. We could avoid the problem in the future by turning off auto-shutdown whenever we chose.

my help in identifying the probHe probably thought really knew my stuff.

Later that day, Wallace acknowledged

lem. "That was a

What

I

good

knew was

call."

I

when

that I'd scared myself almost a year before

I

thought a mistimed auto-shutdown in the middle of the day might ruin sol

...

1

don't

I

couldn't help but think of

come back to haunt you, but to

Even though we hadn't gotten

had asked

fifteen-hour integration.

help you.

the intermediate results the

the one spectrum

for,

And even

all

Sometimes your past mistakes

it.

we

did get

was

APXS team

a clean uninterrupted

We'd bagged our first APXS spectrum of

soil!

engineering analysis team was struggling with the

as the

command

mysteries of auto-shutdown,

we had uploaded

quence to snag Barnacle

In the planning meeting, the Experiment

Bill.

the sol 3

Operations Working Group had been unanimous: Get Barnacle

Bill! It

was the obvious choice. There were no other rocks for yards, Barnacle was almost the ing target. sol 2

right size,

When

through

his

and

its

mottled appearance

made

it

an

se-

Bill

intrigu-

Brian Cooper had surveyed the end-of-day pictures for

3D

goggles, he

was impressed with

his

good

fortune.

The images showed the rover's position at the bottom of the ramp, just after egress. Barnacle Bill

was already so

close

and well aligned with So-

journer that Cooper would not even need to back the rover farther away

from the

lander;

all

Sojourner had to do was turn seventy degrees to

its

APXS to "ramming" position, and drive backwards just about one foot. He checked the rover motions again and again, but they looked right. I merged the traverse commands into a sequence left,

partially

deploy the

that also did a soil mechanics experiment in the

front left

morning (digging the

wheel into the sand) and some imaging.

When

the afternoon results from sol 3

had just had Lots easier

a great day. Sojourner

had

hit

came down, we Barnacle

than our ORTs! The performance of

the best since landing.

And the

rover sent

Bill

realized

on the

the radio

we

first try.

modems was

down its first picture

of the Ian-

Living on Mars Time

285

The image showed mostly

der seen from off-board.

airbags,

but

it

was

beautiful nonetheless.

At

one of the press conferences a reporter asked, "There are those

have looked

Have we

at pictures

and say 'Gee whiz,

landed on Mars, or

really

this

is

this

exist

anywhere on the Earth.

as the arctic. There's

and these

sure,

It's

as

the weather experiment

wouldn't be staying up site

silly

on the

all

Tim

this

if

place.

they had,

would have

it

been faked.

The end of

the rover

more than going

cost

ramp

let

for

"certainly

coming in!"

group of scheming engineers would have designed such an and

as cold

was the lead

was that he

lander. His response

perfect to have

it's

temperature, pres-

Schofield

night to watch fake data

seemed too

"Here's a place that

dry as Arizona and

nowhere on the Earth with

sorts of surface features."

The landing

looks like Arizona

out in the desert somewhere?

Matt Golombek just thought the question was does not

who

out into a large

flat

secret

ideal location,

Mars

to

No

in the first

region nearly de-

void of hazards, just the right spot to get our "Mars legs" and get used to driving

on Mars. Beyond the open

more challenging set

grouped

terrain with a

APXS

target rock, right

site

when

number of

closely together that

Garden." Even better, the

area,

the rover

was

ready,

was

rocks to study, including one

had quickly been dubbed "The Rock

had given us Barnacle

where we wanted

it.

Bill

What more

—the perfect

could the rover

team ask for?

* Over the

first

week, the Data Controllers learned to give the uplink team

Henry Stone seemed more focused on getting the rover status reports complete and on the web than in sifting out those details that were most important to operations. The re-

what

it

needed. At

first,

Allen Sirota and

ports were loaded with data.

how

difficult

it

was

to get everything

Those of us on the uplink care about

all

The Data

the details!

side got

Controllers

done

would complain about

in the limited time available.

more and more

We just wanted the

critical

frustrated.

few

We

facts that

didn't

would

SOJOURNER

286

tell

us whether

something

we

could operate the rover normally, or

special, like

sleep early to prevent overheating.

and the

into a pattern,

Sol

modem

heat the radio

or

tell

to

do

the rover to go to

As the Mars days went

transfer of data

we had

if

by,

we

settled

between team members improved.

5

At the end of

two days

sol 5,

early. In

Sojourner completed

prior sols, Sojourner

tiful

Bill

rock measurements.

image of the Pathfinder

for the

APXS

vealed

On this sol,

lander.

primary mission

had performed

nology experiments, and collected both Barnacle

its

The

soil

objectives,

a full set of tech-

measurements and

Sojourner delivered a beau-

entire lander

was

visible,

except

IMP camera, which rose up out of the field of view. The image reSojourner's own path, as evidenced by wheel tracks leading in an

apparent straight line from the extreme foreground back through the

mottled

The

soil to

rover

the deployed ramp.

team declared

rover extended mission.

success.

We

were now

(We could thank Donna

in the

realm of the

Shirley for negotiating a

practical set of mission success requirements years before.)

But other than

being a nice thing to contemplate in a spare moment, no one noticed.

We

went back to work.

Sol

6

At the latest

sol 6 press conference, Peter

IMP images with

Smith began

he called "a quick observation on the

effects

thirty-nine minutes later every day to stay

When you

say

his presentation

a typically dramatic introduction, a

good morning

as the

sun

of shifting your schedule

on Martian Local

is

of the

poem which

setting,

now

Solar Time":

that's living

on

Martian time.

Your sunglasses look glasses.]

—

like this

that's living

[Smith put on a pair of red /blue

on Martian

time.

3D

Living on Mars Time

Xo

28?

time for laundry and you get your shirts out of the box of proiect

T-shirts even-

day

—

that's living

on Martian

solar time.

When you start admiring strange-looking rocks, talking about

When

them

your friends

to

your days are called

living

laughing

start

They had all

rest

lost all sense

that's

—you know vou're

living a

skewed, warped, intense

exis-

of the normal flow of night and day In the

we were

nearly completely isolated

of the world.

phone messages from

get

on the success of the mission. But lulls in

time.

—

.And Smith had identified a fundamental truth:

it.

of the media attention,

center of

from the

would

on Martian

are called days

engineers' jokes

at the

The mission operations teams were

I

that's living

on Martian time.

The audience loved tence.

—

and your nights

them names and

on Martian time.

But when you living

sols

giving

friends

and

family, congratulations

could never respond, because the few

I

would inevitably occur in

the rush of activity

the early hours of the

morning. Eating was

problem.

a

On

at the right time.

sol

My

1. I

appetite

was back, but

had come on

shift

Time. Lunchtime started out around

later.

Even the pizza places weren't open then. .And dav.

I

was never

quite sure

was never hungry

about 10

light

changing even"

I

2 a.m..

p.m.. Pacific

Day-

and got progressively since

mealtimes were

when mv body would be

ready

to eat.

Sleeping was a problem.

I

might

finally finish

and head out of the Space Flight Operations Center

sunny I

warm day had no time for. I

would

hours.

try to sleep in the

Then

I'd

of those days.

I

the

all

middle of the

day.

MSA. heard Matt

I

found

it.

I

a bright

would succeed

TV

at

for a

work.

few

On one

and surfed through live

feed from

stopped and listened to the voices

Wallace's voice. "Flight Rover."

the attention of the Flight Director. There "Flight. Rover.''

I

was one education channel running the

When

on

drive home, exhausted yet keyed up.

went downstairs, turned on the

the time. I

at 10 a.m..

wake up wondering what was happening

the channels. There

JPL.

I'd

reviewing a sequence

He was

asking for

was no response

he said again. The Flight Director was

in

to him.

listening to

some-

SOJOURNER

288

body else.

"Flight. Rover."

'Answer him already!"

I

As usual, Wallace had more patience than

TV

yelled at the

"I

want

on!" Eventually the Flight Director responded, and

from the dialog

was

It

that the rover

was

a fishbowl existence.

to I

know

could

I

did.

what's going

relax,

knowing

fine.

still

When

else in

my life

would

I

be able to

turn on a television set to watch the minute-by-minute goings on at work?

Somehow, the tarily

project

I

had worked on

for so long

had merged momen-

circulating

around the building.

with the national consciousness.

A few days

after landing, a story

began

Sometime around the second day of phone

call for

surface operations, an urgent tele-

Wes Huntress had come in to JPL. Huntress was the NASA

Associate Administrator for Space Science, head of the office within

NASA that had funded Pathfinder. He was visiting JPL to observe the culmination of the mission. The

Johnson Space Center, the

call

NASA

came from an important personage

at

center that trained the astronauts and

manned program, including all of the Space Shuttle flights. Huntress was tracked down and the JSC official finally got the Associate Administrator on the phone. The official was furious that the JPL operations personnel were presenting the wrong image to the public. Where operated the

were

their shirts

and ties? They were wearing shorts and T-shirts, cheering

and hugging, putting

their feet

control room! All of this

up on counters, even eating pizza

was being transmitted

live to

TV

around the world, and huge numbers of people were seeing the impression of

in the

audiences

it.

Was

this

NASA that the organization wanted to communicate to

Would Huntress please see to it that the Pathfinder team shaped up and conveyed a more professional appearance and attitude?

the public?

Huntress's reported response

was

to

tell

the

JSC

official,

"Fuck

off!"

and

hang up the phone.

Whether or not

the story was true, the Pathfinder operations teams proof of our professionalism was that Pathfinder was on The loved Mars, the lander was functioning perfectly, and Sojourner was driving around up there! And NASA was astute enough to recognize it! And, as the team sensed but which became clearer as time went by, it.

Living on Mars Time

the public faring,

ume day

was

wanted

enthralled. People

traffic

after landing,

know how

to

day by day Hits on the Pathfinder web

of Internet

289

site literally

the rover

was

doubled the vol-

On the Mon-

during the early days of the mission.

when people were coming back to work for the first time

since the Fourth of July holiday, there

were 47 million

hits

on the

site,

as

people downloaded images from Mars. Interest in the mission extended

worldwide. There was a rumor that the entire French telephone / Internet

network had been brought down by the volume of traffic directed at Pathfinder,

and that the French government had asked

its

citizens to

show

re-

straint in order to restore service.

To many of

the people tuning

neers in the control

room looked

in,

the young, casually dressed engi-

a lot like themselves.

Most of the science team, involved only with lander-based instruments,

took rover operations for granted. But whenever Sojourner achieved a

new objective, Hank Moore would send us a congratulatory email:

"Great

job rover team!" The JPL store had started selling Pathfinder baseball caps

with an embroidered Sojourner on one never seemed to take

dered

when

the rover

it off.

Some

would

side.

Once Moore got

his,

he

of the other Pathfinder scientists wonso they could stop factoring rover

die,

IMP

images into their science plans. Moore would grin and reassure the rover uplink engineers: "Sojourner's a tough a year

little girl.

You'll

still

be driving her

from now!"

During breaks

in the action, or if

we were

off shift but afraid to

go

home because we might miss something, some of us would steal into one of the conference rooms to look at the twenty-foot-long poster of the landing

site.

room was a printout of the view from the IMP camera, compos-

Covering an entire wall of the

"Monster Pan," the ited together

full

360-degree

from more than

a

hundred smaller images.

Pathfinder team had hand-lettered

names on

was

It

than to use a

number and

been assigned

to.

The names were

a

about Yogi, Barnacle

Bill,

and Chimp,

remember which rock

the

number had

easier to talk

try to

on the

scraps of Post-it notes and

stuck the notes over various rocks visible in the poster.

convenience:

Scientists

Matt Golombek,

as Pathfinder Project Scientist,

had

SOJOURNER

290

specified rules for rock-naming,

be named

after people.

He

most

particularly that rocks

also implied that since the

the scientists, only the scientists should do the naming. the Pathfinder

team who had any

junction.

Once

knew the

source.

a Post-it note

There were to search out sight.

had

interest in

The engineers on

naming rocks ignored the

unnamed features

in the

It

became

a pastime with the

in-

team

Monster Pan, and correct the over-

Eventually the Post-its were so thick on the poster that

to assign

to

to aid

went up on the poster, no one but the namer

rocks to be named.

still

were not

names were

Golombek

one of the junior members of the team just to keep track of

the names.

Other Pathfinder traditions spontaneously

were one of

these.

arose.

Rover wakeup songs

On sol 3, Howard Eisen had the idea of playing a song

for Sojourner just before the

morning session during which the

sol's

com-

mand sequences would be uplinked to the lander and rover. Mission Control for manned space missions traditionally greeted the astronauts with a wakeup song each morning.

Why not do the same for the rover? The first

song played was "Final Frontier," which sounded appropriate

for a

deep

space mission, but in reality was the love theme from the television series

Mad About

Over the next few

You.

brought in songs to

play,

sols,

rover

and sometimes

team members sometimes

forgot.

But by

sol 12, the habit

was in place. Rover wakeup songs were being played every sol. Not having a

song was no longer acceptable.

act,

playing a

live

sion Support Area.

were

On

one

sol,

solo of "Centurian Starfire"

Some of the media had

Rob Manning got on

his

trumpet

into the

in the Mis-

the impression that the songs

actually being transmitted to Mars. In reality, the rover

wakeup

songs were only sent across the voicenet to be heard by the rover and lander teams on the second floor of Building 230. for

them,

as they

began

a

The songs were

really

new day operating two robotic spacecraft on an-

other world.

The

rover uplink team participated by submitting songs, but almost

never actually heard a wakeup song being played. By the time of the

morning uplink definition, the

session,

we were

already at

wakeup songs greeted

worked the Martian

night.

home,

in bed,

the Martian day.

and

asleep.

By

The uplink team

Living on Mars Time

291

Sol 12

On the night of sol

11,

Sojourner completed an

big rock Yogi. For sol 12, the science

at a

to send the rover to a

named Scooby Doo. On

whitish patch of ground they had

would stop

APXS experiment on the

team asked us

the way,

we

sandy area called the Cabbage Patch to do a wheel abra-

The experiment would cause just

sion technology experiment.

middle wheel to turn, digging

it

were rough enough to scrape

into the soil to see

if

the right

the Martian sands

had been

off the special material that

painted onto the wheel's surface.

Up

traverses using only low-level

hadn't wanted to risk the until

The

stripe

laser stripes

were indeed

visible to

On sol 8, the rover took a complete photograph of one

showed

clearly in the image, stretching into the distance.

team was unanimous

tonomous navigation Peering into the first

of the rover's

"MOVE" and "TURN" commands. We TO WAYPOINT" command

The lasers worked. The Cabbage Patch was only rover

all

autonomous "GO

we'd confirmed that Sojourner's

the rover cameras. laser.

team had accomplished

to this point, the uplink

that

capability

RCW

it

was time

a

few yards away, but the

to take advantage of the au-

we had built into

stereo display, Brian

Sojourner.

Cooper designated the

Martian waypoint, a modest traverse of about seven

commands

feet.

Rick Welch

into the overall sequence,

which

included the wheel abrasion experiment and several rover images.

When

integrated Cooper's traverse

Welch and Cooper came

in to

work the next day,

the engineering analysis report.

The "GO

Sojourner was

sitting in

team gave them the best news possible: nothing much to

TO WAYPOINT" the

had executed without

Cabbage Patch, and the

When

I

first

vehicle

was

had time to stop and

error,

in

good health. What

think,

it

next?

seemed strange

that rover

more smoothly than they ever had on Earth. But it was unquestionably true. Everyone on the rover team recognized it. We were getting great images from the rover cameras. We'd

operations on Mars were going along

been

hitting

our

targets.

(The downlink team had even put together an

APXS target scorecard. A single sheet of paper was posted on the wall outside the engineering analysis area.

tures of each of the rocks the rover

On

it

were the names and small

had reached, reminiscent of the

pictally

SOJOURNER

292

World War II bomber crews painted on the

that

words across the top of the sheet

said,

sides of their aircraft.

The

"You pick 'em. We'll plant 'em.")

Perhaps the relative ease of rover Mars operations wasn't so odd. The

team had trained for months, and the

rover

out in

all

although

failures

during that time stood

our minds. The engineers had learned from

their mistakes.

And

we had dutifully pretended during RORTs and ORTs that Marie

Curie was on Mars, there was no substitute for the sure knowledge that

Sojourner was

now

made

—

team

a mistake

on another planet

truly

to add that

last bit

—

far

beyond rescue

if

we

of care to everything the operations

did.

And there were still problems. Sometimes the rover halted its driving when it encountered a steep slope that wasn't really there. Sojourner's tilt sensors, intended to

were

sticking

why.

If

warn of steep

slopes

and impending rollover hazards,

and sometimes giving erroneous readings.

the Rover Driver

on duty

(either Brian

We didn't know

Cooper or Jack Morrison)

decided the terrain of the rover's next traverse looked

choose to turn off the offending

tilt

safe,

he could

sensor.

Sojourner's turn rate sensor had adopted the unpredictable ways of its

twin on the Earth-bound Marie Curie, sometimes leading the rover

astray as

it

drove. After long traverses, Sojourner might

more from

its

the science

team

end up

a yard or

target. If that occurred, the rover planners consulted

to decide

whether to continue on to the

nation, or go for a nearby "target of opportunity."

We

with

original desti-

were on Mars

Chances were good that some rock close to the rover would

.

.

.

excite the sci-

entists.

Sol

30

Around

sol 30, the rover

had been

living

and lander uplink teams

on Mars time

for a

revolted.

Both teams

month, and we were exhausted. The

adrenaline rush that had kept us functioning during the fourteen-hour shifts for

the

first

two weeks had long

hadn't hit the science teams as hard: stick

since dissipated.

most of the

The schedule

scientists

only had to

around until they had decided what they wanted the lander and rover

Living on Mars Time

to

do during the next

tailed analysis

sol.

Then

293

the scientists could go

of telemetry on their

own

home, or do

Much

schedules.

team had already evaporated. Many of them had returned universities,

forty minutes each day,

to start living

schedules by

—

—so —was far

their rover counter-

command

sequences they

planned were for

IMP

various

specified resolutions. Building a sequence a

ahead was

feasible. If

science imaging

you could do

that,

Was

again?

lander didn't move. Most of the

filters at

friends.

for the foreseeable future.

life

on Earth time

The lander uplink team had one advantage over

The

work

of a wildly successful mission

the prospect of living this Mars-centric

no way

shifting their

still

remained largely isolated from family and

An unexpected consequence

parts:

home

to their

continuing their participation by telephone and Internet.

Meanwhile, the uplink teams,

there

de-

of the science

of pictures taken through

sets

few

sols

you could mostly do the work on

a regular shift.

But Sojourner did not stand

still.

So sequences for a given

not be generated without the telemetry from the previous rover

sol

Yet the

sol.

team was getting better and better, on both the downlink and uplink

Most of the downlink team had

sides.

little

to do, as long as

none of

Sojourner's hardware suddenly failed. In the absence of vehicle alies,

on

could

anom-

the Data Controllers could handle a first-cut downlink analysis

their

own

in a

few hours, alerting the

Matt Wallace came up with a plan:

We

of the team

rest

would

The

schedule, but not completely around the clock. start at 6:00 a.m.,

and the

Data Controllers would

still

Mars afternoon telemetry in the rover data, the

latest

would end by

have to

results,

come

shift

still

necessary.

the uplink team

earliest shift

10:00 p.m.,

in at

if

JPL

would

time.

odd hours, to look

The

at the

but unless there was a major anomaly

Data Controller could go

home

again in a couple of

hours.

With the lander team now working on Earth Rover Coordinators anymore. So

we brought

we didn't need the Thompson and Matt

time,

Art

Wallace onto the uplink team as Rover Drivers /Sequence Planners. That put

six

troller,

engineers on the uplink team.

Howard

putting three engineers in that role.

Report on

sol 32.

Eisen

became

Eisen wrote

a

his first

Data Con-

Downlink

SOJOURNER

294

With an almost

regular

work

schedule, and

trade off responsibilities, the entire

normal

Sol

more team members

team could return

to

to a semblance of a

life.

35

The time had come clustered together,

to visit the

all

Cooper designed the

traverse

—ambitious still

rock aptly called "Wedge," to halt

The next

day,

at

almost twenty-three feet

—to

lander, across a relatively clear patch of

ground, then through a bumpier but like

group of rocks

a

of them big enough to be good APXS targets. Brian

away from the

take Sojourner

Rock Garden. This was

navigable stretch behind a ramp-

at the

Sojourner was not where

Rock Garden

we

"entrance."

expected. Instead of driv-

ing on the far side of the rock Wedge, the rover's dead reckoning error

had led

how

it

in front of

Wedge. At

first

the rover could have gotten to

some of

its

the rover

team wondered

There

seem

position:

didn't

enough room between Wedge and the neighboring rocks drive through.

The

drift in

rounded by rocky

itself

on the wrong

side of

its

thread-the-needle

Wedge and nearby

Wedge,

largely sur-

terrain.

Attempts to leave the area over the next several the rate sensor

its

Sojourner had safely navigated between

Hassock rock, and found

for the rover to

the rate sensor had led the vehicle astray, but

hazard avoidance behavior had been perfect. With

mode enabled,

to be

meant Sojourner often turned

sols failed.

to the

The

drift in

wrong heading, and

the rocks around and under the vehicle tripped safety hazard errors that

aborted the traverses before the rover could get very

far.

Brian Cooper unofficially dubbed the region "The gle."

Bermuda

Trian-

Sojourner had gotten there with ease, but would spend days there

before escaping.

"We had planned

to avoid this entire area, going farther

we circumnavigated the lander and going directly to the Rock So here we had scientists who were very anxious for us to get to

to the left as

Garden.

these rocks which

APXS, and we

had

large vertical surfaces that

were perfect

were spending days and days in this other area."

tration of the uplink

team

increased.

for the

The

frus-

"We were getting nowhere. We'd al-

Living on Mars Time

ternate. Jack

would

out of here. Maybe

try

it

and

would

I

and say 'Maybe

if it

itself at risk.

operators on Earth to correct a problem before

once you were already in rough

terrain, these

it

Wedge

.

.

.

became too

would happen. To him, left for

lucky

the day,

I

this

was the

off."

riskiest

thought there was the

But

would lead it

tried driving to the right

Cooper wasn't sure what

"When I we were un-

time of the mission.

distinct possibility

—

if

—that we would find the rover flipped over the next was capable of much more

day,

and the

Howard

mission would be over." Yet the mechanical team, led by insisted that the rover

severe.

same safeguards could keep

"we eventually

with the safeguards turned

human

to allow the

there, causing the vehicle to abort the very traverse that

to safety So, to get out of there,

of

These protections

encountered too steep a slope, or if the bogie angles

became extreme. The design philosophy had been

you

you'll get us

several safeguards into Sojourner, specif-

prevent the rover from putting

halted the vehicle

it,

will/"

I

The software team had built ically to

try

295

difficult terrain

Eisen,

than the

rover drivers had ever risked. In pure

Bermuda

Triangle tradition, the next downlink revealed So-

journer undamaged but

on top of the

rock,

still

at

Wedge,

and the right

side

this

time with the

left side

wheels

on the ground. The uplink team off." And that's what they Wedge and proceeded to the

chose to keep going: "We'll just drive straight did.

The next

sol,

Sojourner powered over

entrance of the Rock Garden.

The

curse of

now open

Sol

to us.

Wedge was broken. An abundance of APXS targets were Their names were Shark, Half Dome, Moe, and Stimpy.

56

The rover

batteries died in the early

between midnight and

Day Downlink

3 a.m.

than had been promised twice

its

sol 56,

MLST. Henry Stone wrote

Flight Status Report:

FROM HERE ON OUT!"

morning hours of

"WE'RE

in the End-of-

ON A SOLAR

Although the batteries had survived

(at sol

sometime

MISSION far

longer

56 the rover had been operating for almost

specified "extended mission" lifetime), the loss of the batteries

SOJOURNER

296

was

still

The power engineers had been using

a surprise.

to estimate battery usage throughout the mission,

and

rover telemetry their best guess

had been that nearly half the batteries' capacity remained unused. Where had the

rest

of the energy gone? The higher the temperature, the shorter

a battery's shelf

apparently the higher than optimal temperatures

life;

during the cruise to Mars had partially discharged the batteries before the surface mission began.

The power team had had no

data on shelf

the actual temperatures Sojourner had experienced,

now.

And

since the rover

tions, there

at least

had been shut down during APXS night opera-

was no hard data on power consumption

way

overnight periods, no

life at

not until

make

to

for these long

accurate quantitative measurements.

They had guessed low.

How

would the

mission? All future hours.

rover

loss

of Sojourner's batteries impact the rest of her

APXS

And the rover's alarm

would wake with the

By

operations

this

would now occur during

clock would no longer run overnight, so the

sun, or not at

all.

time, the lander too was conserving

der's battery

was being recharged each

degrading, bit by

bit.

So each

sol,

sol

by

its

its

down

battery.

While the

lan-

was

also

solar arrays,

it

although fully charged, the lander bat-

tery stored less total energy than the sol before.

the lander shut

daylight

To reduce power

drain,

completely each night, which meant any science

data stored onboard which had not already been transmitted to Earth

would be

lost forever.

While the lander and rover battery problems were

on the operations team was complaining. them. These were exactly the problems

We

were

we wanted to

serious,

none of us

just dealing with

have, the ones that

showed up only because everything else was working well. The rover and lander were aging, living beyond their design lifetimes.

human beings,

And just

as

with

the difficulties of growing old beat the alternative.

Sojourner had

made

the transition to

its

"solar-only" mission without

incident.

By

sol 59,

we were

calling

telemetry that day, and

it

"The Thompson Loop." Sojourner sent no

we thought we knew why.

Living on Mars Time

29?

command sequences was an involved process. The scientists' requests for mosaicked images and multiple readouts from the APXS forced sequences to be many times longer than we had predicted during Sojourner's design phase. The average sequence was now between two Building

hundred and three hundred commands. Sometimes the Sequence Plan-

would have

ners

memory

part of rover

prune down the commands to

to carefully

a long time to prepare

allocated to hold sequences.

and review. By

this time, the

fit

into the

Long sequences took uplink team consisted

of Cooper, Morrison, Welch, Wallace, Thompson, Laubach, and myself.

We

who thrived on doing what we had not done beSo it was natural that we would find ways to improve the sequences,

were

fore.

all

engineers

and make them

easier to

produce each

had put together templates

would have something

Even before landing, Laubach

week of

operations, so the

team

to start with, instead of building sequences

from

for the first

scratch. Later in the mission,

day's sequence

day.

and modifying

Thompson's sequence

we were it

in the habit of pulling

down for the

was

in a

collection.

Dome

night.

communications blackout to avoid interrupting the data

But Thompson had

tried to eke out too

time: Before 3:30 p.m. came, the sun

rover and its

its

much

data collection

had dropped too low in the sky for So-

journer's solar array alone to provide sufficient

protect

until 3:30 in the after-

Thompson had chosen to specify the Time. As usual when operating the APXS, the

time in Mars Local Solar rover

APXS experiment The rover was commanded

for sol 58 included the first

from the rock dubbed Half

noon, then shut

yester-

for today.

since the exhaustion of Sojourner's batteries.

to collect data

up

power

to sustain both the

APXS. As designed, the rover began shedding power loads

CPU from browning out.

Finally, as its solar

power waned,

to

the

rover could not maintain itself any longer, and like a computer with

plug pulled,

it

stopped operating.

ing sunlight of sol 59,

wakeup time

When the rover came alive in the morn-

assumed the time was

7:15 a.m.

Mars LST, the

that would have been commanded by an auto-shutdown.

Sojourner was the next

it

its

sol. It

was

still still

waiting for 3:30, only in

its

silent

now it was

early

morning of

mode, so there was no way

to send

it

commands. As the operations team reasoned through what was happening,

we worried that

Sojourner might be stuck

in its current state forever,

SOJOURNER

298

always waiting for 3:30, always shutting off before 3:30

completing In

sol,

.

.

.

never

The Thompson Loop.

communications blackout

its

once per

sequence.

its

came

state,

the rover only

communicated

when it woke up in the morning and asked the lander for the

correct time in order to resynchronize the rover's clock. During the time

synch communications session the rover wouldn't even accept

mands. rover

If

we

could only

tell

Sojourner that

it

was almost

3:30,

then the

would complete the wait long before the sun went down, move on

to the next

command, and come out of

the blackout. But

planned on "lying" to the rover about what time

responded to a rover time request with time.

new com-

The only way

its

it

own

was.

we had

never

The lander simply

estimate of the current

to send Sojourner incorrect time

would be

to

change

the lander's clock to a false value, wait for the rover to request the time,

then change the lander's clock back. This would be extremely dangerous for the mission.

der to point

its

The lander depended on knowing

the time of day in or-

high-gain antenna precisely in the direction of the Earth.

Changing the lander clock would point the antenna toward the wrong point in the take the

sky,

disrupting communications. Richard

Cook would not

risk.

On sol 60 rover telemetry came down. The low-voltage condition that had existed just before the rover ceased to function on

sol 58

had triggered

anomalous readings of APXS current, causing the rover to declare the

APXS

a failed device.

On sol 59,

Sojourner had therefore refused to power

on the APXS. Leaving the APXS

off left the rover with

enough power

operate until 3:30 without browning out, complete the

to

command, and

down normally. A sequence was uplinked for sol 60, but Thompson's original sequence had more APXS data collection built into it, so the rover was still in blackout mode, and the new sequence was not received. Would Sojourner stay stuck in the Thompson Loop? On sol 61 the new command sequence got through, and Sojourner exited the Thompshut

son Loop for good.

Thompson have his name as-

Rather than being embarrassed by his starring role in the Loop, Art would only smile. sociated with the incident.

He was

actually

proud to

Living on Mars Time

299

Even driving on Mars can get tedious. Sojourner had been exploring the

Ares

landing

Vallis

site for

operators," recalled

over two months.

"It

was taking

Thompson. "Some people were

ate this vehicle. I'd have to say

I'm driving I

I

come

was among those people,

going to go on?' but then

this

is

this vehicle

on Mars.

I'd

and maybe

it's

I

in to oper-

thinking,

'How

stop and think, 'Wait a minute.

Who else

ever get this opportunity again?' Yeah,

on the

starting to get fa-

tigued and starting to almost complain about having to

long

its toll

can get

gotta get up,

a fourteen-hour, fifteen-hour shift,

opportunity? Will

this I

gotta go to work,

but I'm gonna go look

new set of pictures no one's ever seen before, and I'm gonna load the commands to make this vehicle go somewhere on another planet. That's at a

really great!"

And

Sojourner continued

as

travels,

its

it

began

to peer behind rocks

and capture images of features that even the Pathfinder lander had never

Cooper loved those

seen. Brian

images that came

down were

pictures.

"Some of

dune

the sand

areas

the

more

spectacular

beyond the reach of the

IMP camera, behind the Rock Garden. That was pretty exciting. I wanted to

do

a lot

more of

that."

Cooper looked forward

to later in the mission,

when Sojourner might venture as far as a hundred meters away from the you lander. "We all wanted to see what was behind the next rock in case never know there might be something really cool on the other side of

—

—

that rock

known

.

.

.

besides another rock

.

.

some evidence

.

.

.

.

something un-

..."

Sojourner's days began to

grow

shorter.

It

wasn't that the seasons were

changing. Instead, the opportunities to communicate with the lander were

beginning to shrink. For the early part of the mission, the Deep Space

Network had promised Pathfinder

lots

of coverage from

its

big dish an-

tennas situated strategically around the Earth. But Pathfinder was going

on and losing

on, and there

its priority.

were other customers

And

for the

DSN. Pathfinder was

the lander could only receive messages from the

SOJOURNER

300

DSN when

the Earth

was

visible in the

Martian

As Mars and Earth

sky.

moved in their orbits, the Earth was setting earlier each sol. Sometimes, when everything else was just right, the DSN station on Earth would fail

One of the three DSN stations was located outside of Canberra, Australia. More than once, when no downlink came down from Pathfinder, the word would go out: us for a maddeningly old-fashioned reason: the weather.

"It's

raining in Canberra."

The telemetry

that

had streaked unimpeded

across 120 million miles of interplanetary space instant of

In

its

He

Eisen would grab

often acted as

if

some of

guy on the spot

all

the rover

nearly as crucial a role in

its

the rover

was

when

"his baby,"

that

the time came.

and that no one

else

had

creation as did he. Eisen had often been the

and rover demonstrations, both during

JPL and just prior to launch down

thought he'd maintain the pattern

after landing,

was mostly too busy operating the rover

We

in Florida.

while the rest of the team

much

to spend

time talking

it.

Well, this

few

team had assumed

the media attention

for interviews

rover development at

about

in the last

journey by a mere cloudburst.

the months before landing,

Howard

was defeated

sols.

By

was one more misconception

that did not survive the

first

the time of landing, Eisen had integrated himself into the op-

He seemed to have willingly accepted his role on the

erations team.

engi-

neering analysis team as one of the key representatives of the mobility

and thermal subsystem.

He

regularly provided insightful analysis

and

in-

He simply did a good job. Of course, He continued to readily express strong opinions, and

terpretation of the incoming data.

Eisen was

still

Eisen.

some onto other people's turf. But if he had stopped doing we probably would have wondered after his health Meanwhile, the attention of the media was mostly elsewhere. to tread

.

While

Donna

Bill

Shirley

Layman had embraced

it.

shied

.

away from the Sojourner

limelight,

Just before Pathfinder landed, and then for

weeks afterward, she seemed television

.

that,

to be everywhere.

She appeared on every

network news program and CNN, explaining the Pathfinder

mission and

how

Sojourner came to be. She was widely regarded as the

Living on Mars Time

301

"mother" of Sojourner. Together with the other leaders of the Pathfinder

—Matt Golombek, Rob Manning, Richard Cook, and newly promoted Project Manager Brian Muirhead—Shirley became a celebrity This mission

was an unusual and

role for engineers

thrived. Pathfinder

Tony Spear had

and

scientists,

but

this

group took to

it,

had made them heroes.

instead chosen to fade

few days of the landing. Spear had act as Pathfinder Project

from view, resigning within

identified

two

Manager: either stand up and take the blame

Pathfinder failed, or step aside to

make room

for others if

it

—to take

his place,

if

succeeded.

Pathfinder had succeeded. Spear appointed Brian Muirhead

then the deputy project manager

a

possibilities for his final

—up

until

and then seemingly

disappeared.

The

rover

team had

its

own minor

celebrity.

Well before landing

day,

Brian Cooper had been identified in the press as the Rover Driver, and the rest

of the rover team seemed to disappear from existence. The image of

game enthusiast since childhood, choosing Soamong the treacherous rocks of Ares Vallis, was just too enticing to ignore. In much of the reporting, Cooper was represented as having complete command of the rover; even the role of the science team in selecting targets was lost. And no matter how many times he tried to explain time delay and the complexity of commanding Sojourner, most a lone engineer, video

journer's path

reporters could not get past their

initial

notion of Cooper "joysticking"

the rover through the Martian desert as

model

if it

were

a radio-controlled

car.

The engineers were growing antsy. science

team

more than

a

We had been doing the bidding of the

two months, and had never driven Sojourner much dozen yards away from the lander. As even Matt Golombek for over

acknowledged, "We'd been in the Rock Garden. We'd spent weeks going inches each day, and rocks,

it

was very

frustrating.

They were

and they wanted to go drive ten meters."

to get the chance to to allocate

When

sick of

were we going

go somewhere? The rover team pressed the

some time

scientists

Golombek described the inRover Drivers." Golombek laughed, char-

to long rover traverses.

cident as the "Outburst of the

going to

SOJOURNER

302

acterizing the rover team's attitude:

"'Come

on,

let's

just put the pedal to

the metal/" Matt Wallace led the "insurrection."

The the

was

result

Rock Garden,

would perform

new long-range

a

team would take Sojourner

the rover

several long traverses,

journer's true capabilities. This rover's mobility

sol

84

we

sols in

for a spin.

We

and from these we would learn So-

would

give us the chance to exercise the

and navigation performance over longer distances than

we had ever had time

On

more

rover plan. After a few

for during

didn't hear

from the

In recent sols, the lander pletely overnight.

Its

our Earth-bound testing of Marie Curie.

battery,

lander.

had been commanded

to shut

down com-

although rechargeable unlike Sojourner's,

was known to be near the end of its useful life. That was okay: The lander could operate in a solar-power-only doing, at least for a while. get one

more

mode just

as

Sojourner was already

The lander team had estimated

that they could

night of useful science out of the batteries. For the night of

sol 83, the lander

was instructed

to turn itself

back on

after midnight.

When Pathfinder was silent on the morning of sol 84, the team speculated that the lander's battery night.

CPU

When

the battery failed, so had the lander's electronic clock.

lost track

time of day

had gone dead sometime during the Martian

it

of time. The lander depended on

was and,

therefore,

antenna to reach the Earth. Most ing in the tenna,

wrong

direction.

which operated

carefully pointed.

It

at a

where likely,

its

clock to

in the sky to point

what

tell it

its

high-gain

was

the high-gain antenna

The

point-

But the lander had a separate low-gain an-

much

lower data rate and did not need to be

should be possible to reestablish contact with Path-

finder through that low-gain antenna.

The days went by and

still

we

did not hear

from the

lander.

Sojourner had contingency sequences onboard, designed to kick in

if

the

normal sequence running on the rover terminated and no new sequence arrived

from Earth.

I

had written the

original contingency sequences a

Living on Mars Time

303

year before, about three months before launch.

They had been loaded on-

board before Sojourner was shipped to the Cape.

From a rover-centric would be one in which

point of view, the worst of

all

possible worlds

made a perfect landing, opened its petals, and the rover was dead on arrival. The media reaction would have been 'All that work to get to Mars, all that money, and the rover is just going to sit there forever!" As the sols went by, with the IMP camera operating flawlessly, more and more pictures would come down, documenting the

the Pathfinder lander

rover's continuing inactivity in full color

and

at

higher

and higher resolution. If

the failure had been communications-induced, the original contin-

gency sequences would have avoided that nightmare by causing the rover to begin a fully

autonomous mission

quences would have

commanded

range of about fifteen

at a

a

few days

after landing.

The

se-

the rover to circumnavigate the lander

feet, blindly

searching for

APXS

and rock

soil

sites,

attempting to image the lander, and transmitting the data back in the

hope

that

someone was

still

listening.

As the mission progressed, the relevant,

original contingency plan

and we started to worry that

a

became

ir-

combination of minor glitches

when all we wanted was for the rover to sit still. So, on sol 80, we uplinked a new contingency sequence to replace the old one. Now if we lost touch with Sojourner, it would sit might

where lander,

activate the contingency

it

was

sequence

for six sols, then drive

back toward the

ramp, and airbags were a hazard to the rover,

lander. Since the

we had

created a

keep-out zone around the lander. The rover would try to reach a waypoint located at the center of the lander, while the keep-out zone kept

The combination

from getting too

close.

The

drive in a circle

ten

rover

feet,

would

again and again. This

it

created an unreachable target.

around the

would go on

lander, at a range of about

until the rover received a

command sequence, or encountered a hazard it Four sols later, we were out of touch with the

couldn't handle.

Pathfinder lander.

new

TWENTY-FIVE

WILL BUILD SPACECRAFT FOR FOOD

We

The

never regained contact with the lander or Sojourner.

Pathfinder team attempted to reestablish communications daily

for weeks, then tried again weekly, final failed

with the

and

attempt, the project threw a party

loss

at last

monthly After the

—a Pathfinder wake. Even

of contact on sol 84, Pathfinder had been a success beyond

any of our expectations.

The mission had fore

it

from the media's attention be-

ended. Yet Pathfinder had entered the fabric of the national con-

sciousness.

web

largely disappeared

site

During the three months of surface operations, the Pathfinder

some people claimed that this represented the Internet coming of age. The

received nearly 750 million

unprecedented web old lament "If

we

can send

an updated version:

showed up

traffic

"If

we

hits;

men to the Moon, why can t we can put a skateboard on Mars

in political cartoons,

episode of an animated television

on the comics series.

And

." .

.

." .

.

now had

Sojourner

pages, and even in an

the U.S. Postal Service

sued a stamp commemorating the mission, barely

five

months

is-

after land-

ing day.

Although we were disappointed to see the mission end, the Sojourner

team was already chomping

at the bit to

move on

to

something new.

were designers, developers, and implementers, not maintenance

We

engi-

Will Build

neers.

on the

spoke to us for the

surface of Mars.

rover mission

The

would be wholly

and

far

the Pathfinder lander

we were already planning we could begin work on the

The 2001 rover

By the time

last time,

people to take our places, so

its

305

We were not meant to do the same thing every day, even operating

a robotic vehicle

far.

Spacecraft for Food

was

to

new

to train

next mission.

be more ambitious than Pathfinder by

a different beast.

more complex. Where Sojourner

It

would be bigger,

stayed within about

faster,

fifty feet

lander and drove barely a hundred yards, the rover for 2001 was

tended to move up to

five miles across the

Martian

terrain.

It

would

of in-

dis-

pense with support from the lander altogether, communicating directly

with an orbiter that would swing by overhead for a few minutes twice each

sol.

Rather than merely analyze rocks in place on the surface, the

new rover would drill core samples and retrieve them for eventual return to Earth. And instead of promising a week-long mission, we were committed to operating on Mars for a

full

Earth year.

Most of the Sojourner team moved on as

we began work on

this

next

effort,

to the

we found

than with Sojourner. JPL had been ordered by force

by over one thousand people

capabilities to

NASA to

reduce

its

work-

two years to come. The in-house

machine and produce hardware continued to dwindle

the face of this mandate.

The new rover team was asked

capable vehicle in less time than start

in the

new rover project. And even greater challenges

hoped

more

MFEX, with about the same funds. At the

of Pathfinder, the rover had been just a payload, and

the project had even

to create a

in

to eliminate

it

completely.

dom—both at NASA and JPL—had been that the

some people on

The common

"faster, better,

wis-

cheaper"

Pathfinder mission could not be done; yet the Pathfinder team had always believed in

itself.

With the

success of Pathfinder, the institutional view-

point had transformed again:

now

The new

the centerpiece of the mission.

agement and

at

NASA

It

rover and

seemed

its

instruments were

that everyone in JPL

man-

headquarters believed the ambitious 2001 mission

could be accomplished. Only those of us charged with carrying

it

out

were unsure.

The Pathfinder and Sojourner team members had become victims of their own success. What had been the exception became the rule. Future missions were

now expected to cost even less and do still more.

Pathfinder

SOJOURNER

306

was held up

as the shining

Other projects

one

at

JPL

who had been on

Pathfinder's success

proof that

expectation was realizable.

this

no

talked about "Pathfinderizing" themselves, but

The keys

Pathfinder ever used that term.

—largely the quality of

its

to

people, their freedom to

cut across usual organizational boundaries to solve problems, and the

mandate its

fixed

to

modify the scope of the mission

budget

discussion

—seemed to be lacking

on "How Pathfinder Invented

Washington, D.C., on November

"What would you do again?" Brian

The

trick

5,

differently if

as necessary to

remain within At

in these other efforts. Faster, Better,

1998, the Project

a panel

Cheaper" held

Manager was

you had the chance

to

asked,

do Pathfinder

Muirhead smiled and answered. "That's not the question.

being able to do the same thing again." The environment

is

in

at

JPL had changed. In conversations with the individual Sojourner

was

the mission sal

over,

I

team members

after

encountered a recurring theme, an almost univer-

how unique the rover team had been. For Rick Welch, who had joined the team after launch,

recognition of

his

one regret

was not having been involved from the beginning. "To me, the grave approach just seems fun,

more

fun,

first

of

making the taco or eating

all. I

it? I

mean, building it

satisfying.

.

.

Which is

sort of like doing both.

having the opportunity to actually operate your

would think would be very

.

cradle-to-

Why

own

would you want

JPL was It is

and see the success of

still

the place to be.

".

.

it?" .

And

Welch paused.

A year

after

"It's still

I

how

it

despite the future uncertainties,

You can t compare JPL

the only place that builds robotic spacecraft.

that builds rovers. For the

think

to build

something that you actually couldn't carry right through and see operates,

I

rover creation,

moment. What

else

It is

to anyplace else.

the only place, now,

would you want

to do?"

hard to believe that we're actually on Mars."

Landing Day, the next rover mission was in turmoil. The

re-

quirements had been changed and changed again. Given the combination of limited funds, too

little

time, and ambitious science goals

which we had

not been allowed to compromise, the consensus of the rover team was

Will Build

we

that

Spacecraft for Food

30?

could not deliver what had already been promised by others. For

months, the Mars Exploration Directorate would not accept that message. Or, if

back to

did, the

it

tell

them

engineers in the trenches were not getting any feed-

that the

team did not have the journer team. if

problem was being

momentum

had been so

some of As

the rover team's concerns.

The Directorate knew

the Soit

was

it

was

own independent

set

at least

down on the surface and had

Even

if

each of these elements

to cost less than Pathfinder, there just wasn't

allocated to the

Mars Surveyor Program to put

all

make

ing. JPL

enough money

of these spacecraft:

gether by the 2001 launch date. JPL managers went to ters for relief,

its

of science instruments, and a science /communi-

cations relay orbiter circling the planet.

was made

as

in trou-

originally envisioned, the 2001 mission included the rover with

science payload, the lander that put the rover

to

rover

station.

turned out the Mars Exploration Directorate had heard

It

its

critical to

We weren't a locomotive barreling down the track;

our train kept returning to the

ble.

that

The new

dealt with.

NASA

to-

headquar-

hoping to eliminate or put off some of the 2001 objectives

the mission manageable in scope. Headquarters sent

them

pack-

had pulled off Pathfinder, hadn't they? They had successfully built

a rover before;

why

should the next rover require so

much more money?

Why the sudden attack of conservatism? Headquarters just didn't believe that JPL couldn't

An

do the job they'd signed up

independent review board was convened to determine what the

Directorate could and should be doing.

ager

Tony

board.

on

its

with

for.

Spear,

The

who was no

Former Pathfinder Project Man-

longer working on Mars missions, led the

board's conclusion: Indeed, the 2001 mission had "too

plate"

and could not hope to complete an

full capabilities for

the time and

money

orbiter, lander,

available.

much

and rover

They recom-

mended delaying the 2001 rover to the next launch opportunity, in 2003. And they further advised that the Mars Surveyor Program should focus much more of its resources on its stated objective of returning Martian rock and

soil

samples to the Earth.

So, after six

months of planning and development of

the entire structure of the

Mars Surveyor Program was up

the 2001 rover, for grabs. Study

SOJOURNER

308

teams formed to determine

how best to redesign the program and keep it

within budget. Workshops were set up to give industrial firms the opportunity to suggest alternatives. There

was even talk and political pressure

to

keep a rover on the 2001 mission by refurbishing and flying the Sojourner Marie Curie.

test rover,

To me, amid these seemed

been

to have

uncertainties, the single-mindedness of Pathfinder

lost.

By the

first

anniversary of the day Sojourner Vallis,

how

NASA ceremony was held at JPL,

hon-

and the Pathfinder lander had successfully arrived could so

At

much have

Ares

at

changed?

almost exactly the same time, a

oring those responsible for the major achievements of the previous year. Individuals ects

a variety of JPL activities

and proj-

were being acknowledged, but one of the Pathfinder engineers

next to

saw

and groups associated with

me said,

that

it

was

"It

looks like a Pathfinder reunion!" As

true.

I

hadn't seen so

much

I

sitting

looked around,

I

of the team together in one

place in months.

Chairs had been set up in rows on the mall, under a tent that shaded guests and honorees from the

summer sun. We

faced toward the steps of

JPL administration building. On the first landing had been set a podium, and behind it tables stacked with awards to be handed out. From the

large speakers

came

ence fiction movies.

familiar

music from the soundtracks of several

And everybody was

dressed up!

Ed

JPL

The music ended.

Dr.

Stone, the

Director,

sci-

and Dr. Wes

NASA Associate Administrator for Space Science, presided. told us how much healthier space science was at NASA, com-

Huntress, the

Huntress

pared to early in the decade. Today, JPL had "more missions on the books than ever" in

its

crowd cheered

as

history.

Then he

started listing those missions,

each was named. Finally

it

was time

for the

and the

honorees to

be presented with their awards. As each recipient was named and the specifics

of the award were read aloud, the individual would have the

medal draped around

his or

her neck, and would then have a photograph

taken flanked by Dr. Stone and Dr. Huntress.

Will Build

Among

Spacecraft for Food

members

the other recipients, each of the

core team was awarded the

NASA

Don

Bickler

was the

— and

Service

his

Sojourner baseball

Medal

Hank Moore, wearing

work on

the

I

—was presented with the Exceptional

cap»

award he had been given twentv-one years

we

think

gave

Tom Economou.

even found myself congratulating

'Look

I've

Tom

earlier for

wandered around congratulating each

all

APXS. Matt Golombek turned

laugh.

was

the Viking mission to Mars.

After the ceremony, other.

a suit jacket,

for his participation in the Pathfinder mission. This

nearlv the identical his

his invention of the rocker-

sole recipient of the rare Exceptional Engi-

neering Achievement Medal. .And tie

of the Sojourner

Exceptional Achievement Medal. But

two awards were the most pleasing to me. For bogie,

309

at

you guys! You're

ever seen this before!" a slap

I

to look at the

the terror of

two of us and

started to

actually smiling at each other.

glanced from Matt to

on the back. That's because

this

Tom to

the

is

first

I

don't

Matt, and

time you've

seen us not working together!''

There were buses waiting dence for a luncheon. didn't care.

When we

We

all

which were

tables

bit silly

wearing our medals, but we

arrived at the luncheon, the President of Caltech

shook hands with each of

m

to take us to the Caltech President's Resi-

looked a

us.

We

found another tent

set

up on the lawn,

with white tablecloths. Carafes of iced tea and

lemonade had been placed on each

table.

And

there

was

a buffet

namese spring rolls, salmon, sesame asparagus, chicken, and

The people

sitting at

my

table

had

all

Pathfinder and Sojourner: Allen Sirota. Art Justin Maki. Bill Dias.

Some had with them

of Viet-

fresh melon.

shared the experience of

Thompson. Tom Economou. their spouses or children,

who me

had also experienced the mission, but in a different way. Sitting next to was Sharon Laubach. who

We

were

just

since Pathfinder

engineers

had become

who had done our

my fiancee.

jobs well. But here

we

were, sharing a luncheon with the President of the California Institute ol

Technology and the Director of JPL. were Olvmpians. .And free lunch. All

I

as

someone

could think was

sending probes into deep space,

We

at the table

that, for

this

was

as if

we

commented, here was

the

were wearing medals

as

someone working good

as

it

gets.

at

JPL and

SOJOURNER

310

The group

at the table talked

about the mission.

Tom Economou

wondered whether Sojourner might be operating even now, ing

APXS

spectra and sending

The mission seemed Justin

tea

Maki

them back to

the

mute

collect-

lander.

a very long time ago.

raised his glass for a toast. Glasses of

went up around the

still

table. "Here's to

And that, of course, would be

doing

it

lemonade and iced

again."

the biggest reward of

all.

EPILOGUE

2001

big rover

The missions — launch

mission was rescheduled to 2003. Less grandiose

a lander

window

and an orbiter

in 2001.

—were

begun

Marie Curie was added

to

fill

the open

as a payload to the

2001 Mars lander mission. Meanwhile, the 2003 rover mission was recast as

Mars Sample Return, intended

Then 1998

the

two

"faster, better,

to bring

Mars rocks back to Earth.

cheaper" missions to Mars launched in

— another lander-orbiter pair—

failed.

The combined

missions had been less than the single Pathfinder mission.

cost of the

ministrator admitted that "faster, better, cheaper" had gone too In the reassessment that followed, the 2001

cancelled,

was

and Marie Curie went back into

also eliminated.

It

two

The XASA Adfar.

Mars lander mission was

a box.

Mars Sample Return

appeared that no mission to Mars would launch

during the 2003 opportunity.

Then

a

few Pathfinder alumni proposed

the ashes of the others.

Bv combining

the original 2001 big rover,

it

might

JPL: Build not one. but distinct landing sites

two

new

mission to

rise

from

the Pathfinder lander design with

be possible to deliver

NASA NASA headquarters

rover to Mars by early 2004.

Rover mission was born.

still

a

a

long-range

agreed, and the Mars Exploration

asked one more thing of

identical rover spacecraft, to be sent to

and operated simultaneously.

two

SOJOURNER

312

On

Columbia disintegrated

dur-

ing reentry over Texas, killing the seven astronauts onboard. Despite

new

February

1,

2003, the Space Shuttle

uncertainties over the future of the space

program

in the tragedy's after-

math, two rovers, direct descendants of Sojourner and Pathfinder, are on their

way to

Mars.

GLOSSARY

ACS

Attitude Control System

ADM

APXS Deployment Mechanism

Al

Artificial Intelligence

APXS

.Alpha Proton X-ray Spectrometer

ATLO

Assembly

Caltech

California Institute of

CARD

Computer- Aided Remote Driving

CCD

Charge-Coupled Device;

CMU

Carnegie-Mellon University

CPU

Central Processing Unit

CRT

Cathode Ray Tube; picture tube

DARPA

Defense Advance Research Projects Agency

DDF

Director's Discretionary

DSN

Deep Space Network

EDL

Entry, Descent,

Test,

and Launch Operations

Technology

solid-state

Fund

and Landing

camera

314

Glossary

Permanent Read-Only Memory

EEPROM

Electrically Eraseable

EOWG

Experiment Operations Working Group

ETL

U.S.

FET

Field Effect Transistor

FUR

Flight Unit Rover; also

g

1

GALCIT

Army Engineer Topographic Laboratories

known

as

Sojourner

Earth gravity

Guggenheim Aeronautical

Laboratory, California Institute of Tech-

nology

GM

General Motors

GSE

Ground Support Equipment

HGA

High-Gain Antenna

IMP

Imager

JATO

Jet- Assisted

J PL

Jet Propulsion

KSC

Kennedy Space Center

MESUR

Mars Environmental SURvey

MFEX

Microrover Flight Experiment

MLST

Mars Local Solar Time

MRSR

Mars Rover Sample Return

MSA

Mission Support Area; Mission Control for Pathfinder operations

MSM

Mars Science Microrover

NASA

National Aeronautics and Space Administration

NATO

North

ORT

Operations Readiness Test

RCW

Rover Control Workstation

RHU

Radioisotope Heater Unit

RORT

Rover Operations Readiness Test

for

Mars Pathfinder Take-Off rocket Laboratory

Atlantic Treaty Orgranization

Glossary

RTTV

Robotic Technology Test Vehicle

SAN

Semi- Autonomous Navigation

SDM

Software Development Model rover

SGI

Silicon Graphics Incorporated

SIM

System Integration Model:

SLIM

Surface Lander Investigation of Mars

SLRV

Survevor Lunar Roving \ ehicle

TACOM

ULS.

USGS

United States Geological Survey

VOCA

\ bice

WEB

Warm

also

known

315

as the

Army Tank Automotive Command

Operated Communications Assembly Electronics

Box

Marie Curie rover

DRAMATIS PERSONAE

Don Bickler

JPL mechanical engineer; inventor of the rocker-bogie; member of Sojourner operations team

Gary Bolotin

Rover lead electronics engineer

Brian Cooper

Software architect and developer of the Rover Control Workstation; Rover Driver during Mars operations

Howard Eisen

Leader of the rover mechanical, thermal, and mobility team;

member of the

rover downlink

team during Mars

operations Matt Golombek

Mars Pathfinder Project

Ken Jewett

Rover mechanical engineer; responsible for Sojourner's

Scientist

overall configuration

Arthur "Lonne" Lane

Task manager of the Mars Science Microrover demonstration

Sharon Laubach

Caltech graduate student; tions

Bill

Layman

Jake Matijevic

member of Sojourner opera-

team

Rover Chief Engineer Rover lead system engineer; promoted to Sojourner

team leader

Dramatis Personae

David Miller

JPL

31?

robotics engineer; early proponent of the micro-

rover concept

Andrew Mishkin

Rover system engineer; Sojourner Sequence Planner

Jack Morrison

Lead software

architect

and developer

for Sojourner's

onboard software; Rover Driver during Mars operations Tarn

Nguyen

Sojourner software engineer

Jim Parkyn

JPL communications engineer

Glenn Reeves

Lander lead software engineer

Carl Ruoff

Supervisor of the JPL Robotics group

Donna

Sojourner team leader; promoted to director of the

Shirley

Mars Exploration Program Allen Sirota

Rover system engineer; integration and neer; Data Controller during

test lead engi-

Mars operations

Manager

Tony Spear

Pathfinder Project

Henry Stone

Leader of the rover control and navigation team; Data Controller during Mars operations

Scot Stride Lin

Sukamto

(Lin

van Nieuwstadt)

Jan Tarsala Art

Thompson

Rover communications engineer Leader of the Sojourner telecommunications

subsystem team during rover development

Rover communications engineer during Mars operations Rover system engineer; Rover Coordinator during operations

Matt Wallace

Rover power subsystem engineer; Rover Coordinator during operations

Rick Welch

Sojourner Sequence Planner and Rover Driver

AlWen

Lead Sojourner thermal engineer

Bob Wilson

System engineer

for the

Mars Science Microrover

demonstration Brian Wilcox

Supervisor of the JPL Robotic Vehicles group; inventor

of

CARD; member of Sojourner operations team

ACKNOWLEDGMENTS

When

work on this book, I imagined it to be a solitary effort. By now it has become clear just how many people have contributed to its sucI

started

cess. Significant

accomplishment

is

rarely achieved

ing alone. As with the development of Sojourner looks, the I

am

more

by one individual itself,

act-

the deeper one

contributors one finds.

indebted to the

many

rover

team members who provided

their

recollections or reviewed portions of the text for accuracy including Brian

Wilcox, Henry Stone, Matt Wallace, Allen Sirota, Brian Cooper, Art

Thompson, Jack Morrison, Ken

Jewett, Scot Stride,

Bill

Don

Eisen, Jan Tarsala,

Bickler,

Howard

Bolotin, Rick

Sharon Laubach, Hank Moore, Joy

Welch, Jake Matijevic,

Dias,

Gary

and

Bill

Layman. Other

Crisp,

partici-

pants in the Pathfinder mission or the rover-related history herein also

al-

lowed themselves to be subjected to interviews: Bob Anderson, Justin Maki, David Gruel, Matthew Golombek, Arthur

L.

Lane, Robert Wilson,

and Ken Manatt. Friends, associates,

feedback:

Bill

and

relatives read drafts

manageable length.

me

important

Deborah Bass, Mark Adler, Andy Mary Forgione provided special editorial originally voluminous manuscript to a more

Hicks, Jan Ludwinski,

Morrison, and Bobbie Laubach. assistance in reducing the

and gave

Acknowledgments

319

Several people at JPL helped in the process of tracking

graphs, dealing with permission issues, and

down

making the photos

photo-

available.

Sue LaVoie, Xaviant Ford, Grace Fisher- Adams, Michael Jameson, Jeanne

Rademacher,

Tom Thaller,

and David Deats each provided invaluable

as-

sistance. I

vice,

offer thanks to

and

faith in the

my

agent,

Agnes Birnbaum,

book; and to

my

for her persistence, ad-

editor Natalee Rosenstein at

The

Berkley Publishing Group. Esther Strauss at Berkley always responded cheerfully to

ous

my seemingly unending stream of questions regarding vari-

details

of production.

want

particularly to

I

thank

my wife,

the long hours of writing and editing, pletely into another world. I've

Sharon, for her patience during

when

come back!

I

often disappeared com-

INDEX Page numbers in

ACS

italics indicate illustrations; those in bold indicate tables.

(Altitude Control System), 224, 225-27, 229

ADM.

See

Army Tank Automotive Command (TACOM),

APXS Deployment Mechanism

33,

48-49

37,

'Aerobraking," 261

Arroyo Seco,

Airbags, 147, 266

Asmar, Sami, 263, 272

Alahuzos, George, 128-29, 131, 241, 242

Assembly, Test, and Launch Operations (ATLO),

Alkalaj, Leon, 91

landing on Mars (July

on Mars

4, 1997),

Asteroid belt,

277

time, 282-83, 286, 295, 297

ATLO

(Assembly, Test, and Launch Operations), 159, 162

Atmosphere of Mars,

Pathfinder, 71, 72

Microrover Flight Experiment (MFEX),

93, 95,

1

xi,

146

'Autonomous Vehicle Research atJPL," 50-51 Auto-shutdown of

176-77

rover, 209-10,

258-59

Altitude Control System (ACS), 224, 225-27, 229

Barnacle

Bill (rock),

275-76, 284, 285, 286

Battery problems, 295-96, 302

Beaudette, Charley, 37

'Amboys, The," 243

"Behavior control," 62, 69, 76

Ames

Bekker, M. G. "Greg," 16, 26

Research Center, 65

Antarctic meteorite (ALH84001), 207

"Bermuda

Apollo spacecraft, 146

Bickler,

Apple Computer, 28

APXS.

See

Alpha Proton X-ray Spectrometer

APXS Deployment Mechanism (ADM) centrifuge testing damage, 161-62, 163 software, 191, 193 testing,

283-84

76-84

sensor head, non-functioning. 212 testing,

ix

Astro Aerospace, 148, 149, 150

Mars Environmental SURvey (MESUR)

noise filtering,

76

159, 162

Alpha Proton X-ray Spectrometer (APXS)

living

9, 29, 46, 48, 55, 75,

249

Triangle, The," 294-95

Don

Exceptional Engineering Achievement Medal,

309

Mars Environmental SURvey (MESUR) Pathfinder, 69

Mars Rover Sample Return (MRSR),

AresVallis, 2-3,202, 299

mobility optimization of rovers,

Army

team

Engineer Topographic Laboratories (ETL), 31

55, 57-59,

60 1

3-24

building, 90

tethered

vs.

untethered rover, 104-5

322

Index

"Blue Rover," 15-17

"Common enemy"

Bogie suspension, 17-18, 19-20, 20-21

Communications, 165-74

Bolotin, Gary, 125-26, 127

paradigm, 244-45

charge buildup prevention, 166

Braun, Dave, 113

crystal oscillators of radios, 170-71

Brooks, Rodney, 61-62

frequency

Bumps

vs. steps,

23-24, 57-58

of radios, 171-72, 173,

drift issue

268, 269-70, 278-79

lander electronics board problem, 168, 172 California Institute of

Cameras on

CARD.

Technology (Caltech),

8,

9

rover, 135-39

Cardone.John, 115-16

Motorola

RNET 9600 radio modem,

60,

105

166

radios of rover as commercial product, 169-70,

195

CCD (Charge-Coupled Device),

136-37, 138

Wood Elementary School,

203-4, 206

Central Processing Unit (CPU)

Mode"

"Surface Operations

Microrover Flight Experiment (MFEX), 111

25-37 "frame grabbers," 34-35

tethered vs. untethered rover, 98, 99

launching rockets from

APXS Deployment Mechanism (ADM) damage from, 161-62, 163

levels,

31-33,37

Rover Gladiators on the Moon, 2 "stereo waypoint designation" system, 32-37

157

Rover (FUR) "Sojourner,"

157, 160,

three-dimensional viewer, 34-35

video-impaired and time-delayed rover, 27-28

161, 189

Ground Support Equipment (GSE), six- axis test,

Pathfinder, 69-70, 78, 79-80

Rover Control Workstation (RCW), 198

159, 162

Field-Effect Transistors (FETs), 162, 163

"H-bridge"

25

robotic reconnaissance vehicles for military, 30,

Assembly, Test, and Launch Operations

Flight Unit

aircraft,

Mars Environmental SURvey (MESUR)

Centrifuge testing, 155-64

acceptance"

166-67

Computer-Aided Remote Driving (CARD),

power constraint and, 91-92

(ATLO),

test,

temperature impact on radios, 171-72, 269-70,

278-79

Mars Rover Sample Return (MRSR), 53

"flight

170-71

NASA,

planetary protection policy of

Carnegie-Mellon University (CMU), 58-59,

Cedar

266-73, 277,

difficulties,

278-79

Computer-Aided Remote Driving

See

landing on Mars,

circuits, 162,

162

wheels for rover, 30-3 See also Semi- Autonomous Navigation

163

(SAN)

186, 187, 255, 302-3

Contingency scenarios,

157

Software Development Model (SDM), 158

Contractors on Rover Team, xvi

System Integration Model (SIM) "Marie Curie,"

Control issues of rovers. See Communications;

Computer- Aided Remote Driving (CARD);

155, 156-60

wakeup of rover, 160-61, 163 Wyle Labs for, 156-57, 158, 162 Challenger

disaster, 11,

Noise tion

State),

202-3

cruisin' to

Mars, 229

on Mars time,

Charge-Coupled Device (CCD), 136-37, 138

living

Charon, x

testing, 251, 255,

Code R (NASA) funding,

80-81, 100, 101, 102-3,

S

(NASA) funding,

Cognizant Engineers,

Columbia

disaster,

100, 103

Cooper, Brian

87, 90, 93-94,

living

107-8

on Mars time,

4, 1997),

275, 276

284, 291, 294, 295, 297,

299, 301

Mars Environmental SURvey (MESUR)

312

Comets, x

Pathfinder, 69-70, 77

Command sequences living

298, 301

259

landing on Mars (July

105

Code

Semi-Autonomous Naviga-

Cook, Richard

96

Channeled Scabland (Washington

filtering;

(SAN)

on Mars

operations team, 242

Rover Control Workstation (RCW), 196-97,

time, 297

operations team, 234, 235, 236-37, 238

Rover Control Workstation (RCW), 197-98, 199,201,202

195,

198, 199, 201, 202, 203-4, 205, 206

rovers (Marie Curie, Software

Model, Sojourner), 129

Development

1

Index

Semi-Autonomous Navigation (SAN), 45-47,

landing on Mars (July

49,50 testing,

CPU.

323

living

on Mars time,

4, 1997),

275

290, 293, 295, 300

Mars Rover Sample Return (MRSR), 57 Microrover Flight Experiment (MFEX),

251-52

See Central Processing Unit

Cruisin' to Mars, 223-3

109,

111-12, 118, 120, 121, 122, 153

Altitude Control System (ACS), 224, 225-27, 229

Rover Control Workstation (RCW), 201

orientation of spacecraft, 223, 224

software, 191, 193

patches for Attitude Control System (ACS), up-

team building, 87-88

loading, 225-27

255

testing, 251,

Sojourner healthcheck, 227-31

Emblem of MFEX, 153-54 "END-OF-DAY" time, 283

spin rate of spacecraft, 223, 224

Engineering Model electronics boards, 127-28,

radio signals, weak, 226-27

sun sensor problems, 223-25

129-31

Crystal oscillators of radios, 170-71, 269

Engineer Topographic Laboratories (ETL), 31

Curie, Marie, 133

Entry, Descent,

"Cussedness of Inanimate Objects, The" (Moore),

257-59

Error messages, troubleshooting guide, 194

ETL

DARPA (Defense Advanced

Research Projects

(Engineer Topographic Laboratories), 31

Experiment Operations Working Group meeting, 235-36

Agency), 31

Data Controllers, 234, 235, 285-86

DDF (Director's Discretionary Fund), Deep Space Network (DSN),

96, 236,

Explorer

40-41

1

,

9

Extraterrestrial vehicles (planetary rovers), 14-15

299-300

Defense Advanced Research Projects Agency

(DARPA), 31 Deimos,

Landing (EDL), 246-49, 261-62,

264-65

Fake landing questions from press, 285 "Faster, better, cheaper," xiii-xiv, 12, 92, 100, 122,

305-6,311

xi

Delta rocket, 220, 221-22

Female gender of

Desai, Rajiv, 79, 89

Field-Effect Transistors (FETs), 162, 163

"Designing on the path of least regret," 110

Field-testing, 204-5,

Dias,

Bill,

rover, 132

240-43

"Flex-cable" technology, 115-18

255, 273

Director's Discretionary

Fund (DDF), 40-41

Flight electronics fabrication, 127-28, 129-30

"Donna's Rule," 108

Flight microrover, 81

Downlink team,

Flight Unit

234, 235

"Driver's test" for Brian Cooper, 204

Driving the rover. See Living on Mars time; Operations team;

Rover Control Workstation

Earth,

132, 157, 160,

Sojourner

Flipping danger, 67-68, 149-50

Frequency

(RCW); Testing

DSN (Deep Space Network),

Rover (FUR) "Sojourner,"

161, 189. See also

drift issue

of radios, 171-72, 173, 268,

269-70, 278-79 96, 236,

299-300

FUR.

See Flight Unit

Rover

Galileo, 10, 11-12,281

ix, x, x, xi

Earth's

Moon,

Earth's

motion impact on launch, 218

Gas

x, xi

Economou, Thanasis "Tom"

giants, ix

General Motors,

Gennery Don,

25, 26, 28

48, 60-61

job security, 212

Go-For

noise filtering, 177, 178, 179, 180, 181, 182, 183

"Go-For," forkwheeled mobility system, 67-68, 69

EDL (Entry,

Descent, Landing), 246-49, 261-62,

264-65

80C85 microprocessor, 91-92, Eisen,

2,

203

Golombek, Matt lander, 145

126, 188

Howard

centrifuge testing, 157, 162, 163-64

on Mars time, 285, 289-90, 301-2 Mars Environmental SURvey (MESUR)

living

Pathfinder, 66, 67, 68, 71, 72-73

job security, 211-12

noise filtering, 183

lander, 147, 148, 149, 150

Rover Control Workstation (RCW), 202

324

Index

"GO TO WAYPOINT" command, Groundwork.

'Autonomous Vehicle Research

140-41, 291

Computer- Aided Remote Driving

See

(CARD); Jet Propulsion Laboratory

Cassini mission,

Electronics and Control division, 14, 50

Mars Rover Sample Return

(MRSR); Mobility optimization of

50-51

charge numbers for projects, 38

(JPL);

Mars Environmental SURvey (MESUR) Pathfinder;

at JPL,"

1

facilities of, 7-8, 15

rovers;

cheaper,"

"faster, better,

Semi- Autonomous Navigation (SAN)

xiii-xiv, 12, 92, 100,

122,305-6,311

Gruel, Dave "Gremlin," 244, 251, 252, 253, 254,

129-30

flight electronics fabrication, 127,

Galileo, 10, 11-12,281

255, 259

Guggenheim Aeronautical Laboratory of

Grand Tour of the outer planets,

the Cali-

Technology (GALCIT),

fornia Institute of

high-profile

8,9

Jet Assisted

10

problems and, 11-12 Take-Off (JATOs), 9

Langley Research Center and, 10

Hanson, Joe, 33,36, 37

Hardware

Magellan mission, 10-11, 12

vs. software, 191-92,

193-94

Mariner missions, 10

Hazard detection system Imager

for

Mars Exploration Directorate,

Mars Pathfinder (IMP),

Mars Observer,

135, 136,

140-44 living

307

NASA ceremony honoring major achieve-

294, 295

ments, 308-10

software and, 186

and Instruments, 66

Office of Space Science

circuits, 162, 163

origins of, 8-9

"Heliopause," 7

Hickey Greg, 113

Robotics groups,

Hot Wheels toy

rover,

Hubbard,

71-72

Scott,

255-56

75-76, 89-90

solar cell production, 13

11

Company

Aircraft

14, 15, 29, 39,

rocket propulsion, 9

Rover Team, xv-xvi

Hubble Space Telescope,

Hughes

122, 208,

96

Mechanical Systems division, 14

on Mars time,

"H-bridge"

11, 12,

10,

Space business and,

102

2, 8,

306

space probes, 9-1

Huntress, Wes, 288, 308

Surveyors, 10, 15

IDM

(Instrument Deployment Mechanism), 104,

"tradespaces" evaluation, 21

United States budget impact on,

105 "Igloo tunnel," 115, 116

Imager

for

Mars Pathfinder (IMP), 134-44

autonomous

traverse

Voyagers

rover,

1

World War

command, 140-41

brain of rover as camera, 136-39

cameras on

1

Viking, 10, 146

See also

and II

2, 7, 10,

for,

9

Computer- Aided Remote Driving

(CARD); Launch

135-39

281

and funding

to Mars;

Mars Environ-

Charge-Coupled Device (CCD), 136-37, 138

mental SURvey (MESUR) Pathfinder; Mars

"GO TO WAYPOINT" command,

Rover Sample Return (MRSR); Mobility op-

140-41

hazard detection system, 135, 136, 140-44

timization of rovers; National Aeronautics

landing on Mars (July

and Space Administration (NASA); Rovers

4,

1997),

265-66

"stripe projector," 140

(Marie Curie, Software Development

"structured light" approach, 135, 139-40

Model, Sojourner); Semi-Autonomous Nav-

temperature of Mars and visual noise, 138, 139 testing, 248-49,

Inner planets,

ix,

Job

security,

207-13

Alpha Proton X-ray Spectrometer (APXS) sen-

"Insurance panorama," 248

sor head, non-functioning, 212

wind, 7

auto-shutdown of Jet Propulsion

Army

Laboratory (JPL), 7-12

Ballistic Missile

Team building

Jewett, Ken, 118-20, 211-13

x

Instrument Deployment Mechanism (IDM), 104, 105

Interstellar

igation (SAN); Jetty Park, 219-20

250

Agency partner with,

Automation and Control, 29

final

9

rover, 209-10,

Kennedy Space Center, life

283-84

assembly of rover, 212

on Mars,

209, 211, 212

possibility, xii,

207-8

1

Index

WAIT' command,

"Local-Time

325

209-1

Imager

meteorite (ALH84001), 207

midnight

error,

210-11

lockup, 264

media coverage of

prelaunch party; 213 "sarcophagus" for Sojourner

silent rover,

271-72

press conference, 279-80

airline travel,

211-12 stress

Mars Pathfinder (IMP), 265-66

for

lithobraking, 262

rover deployment, 274-77

of team, 208-9

Rover Navigation Imaging (movies), 273-74, 276, 277-78

success, dedication to, 213

Johnson Space Center ( JSC) worry about image of

mode" of rover, 270

"silent

Sojourner health check, 264

operations teams, 288 Jovian system of moons, 10

stand-up of rover, 274-75, 277

JPL. See Jet Propulsion Laboratory

temperature impact on radios, 171-72, 269-70.

Jupiter,

278-79

ix, x, 10, 11

Lane, Arthur "Lonne"

Mars Environmental SURvey (MESUR)

Kan, Ed, 1,2

Katzmann, Steven,

47, 49

Kennedy Space Center,

Pathfinder, 68-69, 70-71, 74-75, 77, 78-79, 80

209, 211, 212

tethered vs. untethered rover, 99-100, 101-3,

"Killing the problem," 110, 171

104, 105

Laubach, Sharon landing on Mars (July

Lander, 145-54 "aeroshell" protection, 146 airbags, 147,

266

operations team, 238

Launch

Astro Aerospace, 148, 149, 150

communications stopping from, 302-3 delays

Mars (December

to

Earth's

219-20

Jetty Park,

flipping danger, 149-50

launch windows, 218 Mars's orbit impact on, 218

heatshield, 146

"MODEM_POWER_B" command, NASA review of rover, 152 for,

planet alignments

167

for,

217

prelaunch party, 219

weather impact on, 217, 218-19

146

petals of, 147, 248, 252

Landing on Mars:

See also Cruisin' to Mars;

"powered descent," 145-46

ramps

217-22

motion impact on, 218

electronics board problem, 168, 172

parachute

4, 1996),

Delta rocket, 220, 221-22

164

in,

1997), 263, 278, 280

4,

on Mars time, 297

living

ing

Layman,

Bill

rover impact, minimizing, 150-52

"designing on the path of least regret," 110

rover team

Imager for Mars Pathfinder (IMP), 135-36,

vs.,

2AA-A5

self-righting, 147

problem," 110, 171

"killing the

tetrahedron (four-sided pyramid), 146-A7

landing on Mars (July

Landing on Mars (July

4, 1997),

261-81

living

1997), 280-81

4,

on Mars time, 300

Microrover Flight Experiment (MFEX), 108-9,

"aerobraking," 261 airbag assessment panorama, 266

110, 111, 120,

121-22

Alpha Proton X-ray Spectrometer (APXS), 277

Rover Control Workstation (RCW), 201

AresVallis, 2-3,202, 299

rovers (Marie Curie, Software

Bill (rock),

communications Entry, Descent,

frequency

138,

139

separation connector, 150-52

Barnacle

Liv-

time; Operations team; Rover

missions (new); Testing

for rover, 147-50, 249, 252-53, 255

"reed-relay" 151

on Mars

difficulties,

team

266-73, 277, 278-79

Landing (EDL), 261-62, 264-65

drift issue

Development

Model, Sojourner^. 130-31

275-76

of radios, 171-72, 173,

268, 269-70, 278-79

building, 88-89, 91. 94

tethered

vs.

untethered rover, 97, 98, 99

Length of mission,

93,

256-57

Lewis Research Center. \\

on Mars

"HAL9000," 278

Life

high-gain antenna, 265, 270-71

Lithobraking, 262

possibility,

i

xii.

207-8

326

Index

Living on Mars time, 282-303, 285

"Local-Time

Alpha Proton X-ray Spectrometer (APXS),

WAIT" command,

209-1

Loch, John, 62

282-83, 286, 295, 297

auto-shutdown of Barnacle

rover, 209-10,

Bill (rock),

283-84

284, 285, 286

Magellan mission, 10-11, 12 Maki, Justin, 273-74, 276, 277 Malina, Frank J., 8-9

baseball caps for team, 289

battery problems, 295-96, 302

Manatt, Ken, 75, 79-80

"Bermuda

Manning, Rob

Triangle, The," 294-95

Cabbage Patch, 291

landing on Mars (July

command sequences,

building, 297

on Mars time,

living

contingency sequences of rover, 302-3

1997), 261-62, 263

4,

290, 301

testing, 246-47, 251

Data Controllers, 285-86

Marie Curie. See System Integration Model

Deep Space Network (DSN), 299-300 "END-OF-DAY" time, 283

Mariner missions, 10

from

fake landing questions

press, 285

"GO TO WAYPOINT" command,

291

Mars atmosphere

of, xi,

146

channels, xi

hazard detection system, 294, 295

Earth, distance from, xi

Johnson Space Center (JSC) worry about

gravity, xi

image of operations teams, 288 communications stopping, 302-3

lander,

landing

site poster,

media coverage

of,

life

on, possibility of,

moons

289-90

polar ice caps

287-88, 300-301

"sol" (Martian day),

of, xi xi,

209-11, 234-35

Monster Pan, 289-90

solar system and, ix-xii,

"Outburst of the Rover Drivers," 301-2

surface area of, xi

poem on

Martian Local Solar Time (Smith),

286-87

temperature variances

winds on,

press conferences, 285, 286-87

professionalism of operations teams, 288 public interest

in,

288-89, 304

Rock Garden, 294-95,

299, 301

207-8

xii,

of, xi

x

of, xi

xi

Mars Environmental SURvey (MESUR)

Alpha Proton X-ray Spectrometer (APXS),

Ames

rover team revolt, 292-94

"behavior control," 69, 76

"silent

difficulties,

286-88

mode" of rover, 297

71,

72

rock-naming, 289-90

scheduling

Pathfinder,

65-81

Research Center, 65

canceling

Code

of,

145

R (NASA) funding,

80-81, 100, 101,

sol 3,

282-85

sol 5,

286

Code

sol 6,

286-87

Computer-Aided Remote Driving (CARD),

102-3, 105 S

(NASA) funding,

100, 103

sol 12,

291-92

sol 30,

292-94

computing power,

sol 35,

294-95

cost estimate, 65, 145

sol 56,

295-96

demonstration of Mars Science Microrover

sol 59,

296-98

sol 84, 302,

69-70, 78, 79-80 location, 67, 68-69

(MSM), 74-80

304

flight microrover, 81

solar-only mission, 295-96

flipping danger, 67-68

success requirements, met, 286

funding

Thompson Loop, 296-98

"Go-For," forkwheeled mobility system, 67-68,

tilt /turn

rate sensors problems, 240, 292

of, 96,

100

69

MSM,

uplink team, 285-86, 292-94, 297

lander for

wakeup songs

Mars Science Microrover (MSM) "Rocky

Wedge

for rover, 290

(rock), 294, 295

wheel abrasion experiment, 291

70 4,"

66-67,68-71,72-80,81 micro-devices, 66

Lloyd, Jim, 128,241

^

7

1

1

Index

"self-righting" rovers,

Sixth

67-68

stereo cameras, 54

Mars Science Working Group, 71-74

Surface Lander Investigation of Mars (SLIM), 71-72, 73-74

Imager

See also for

Mars Pathfinder (IMP); LanExperiment (MFEX);

der; Microrover Flight

Rovers (Marie Curie, Software Develop-

ment Model, Tethered

trade studies, 56

walking vehicles, 59-60 weight problems,

"Tooth," 62, 73 See also

32?

vs.

Sojourner);

Team

Pathfinder

Mars Sample Return,

3

1

Mars Science Microrover (MSM) "Rocky Mars's orbit impact on launch, 218

122, 208, 307

Mars Surveyor Program, 307-8

Mars Exploration Rover mission, 311-12

Martin Marietta Aerospace Company, 10

Mars Observer,

Marzwell, Neville,

11, 12,

96

Mars Pathfinder Mission.

4," 66-67,

68-71,72-80,81

building;

untethered rover

Mars Exploration Directorate,

54, 59, 65

Mars Environmental SURvey (MESUR)

See Jet Propulsion Labora-

1

Massachusetts Institute of Technology (MIT), 61-62

tory (JPL); Lander; Launch to Mars; Mars

Mass

Environmental SURvey (MESUR)

Matijevic, Jake

constraints, 85, 86

Pathfinder; Rovers (Marie Curie, Software

centrifuge testing, 159

Development Model, Sojourner); Team

landing on Mars (July

building

Microrover Flight Experiment (MFEX), 122

Mars Rover Sample Return (MRSR), 52-64

4, 1997),

266, 271-72, 280

operations team, 233, 235, 236

air-conditioning unit, 53

Rover Control Workstation (RCW), 200, 201, 206

ambitious plan

software, 190, 191

for,

56

Ambler, 59-60

bumps

vs. steps,

Matthies, Larry, 61

Max

57-58

Carnegie-Mellon University

(CMU)

Field Ro-

Planck Chemical

Media attention

botics Center, 58-59, 60

on Mars

living

computing power, on

silent rover,

vehicle, 52-53

cost estimate, 63-64

"encoders" for measuring

tilts

of compart-

time, 287-88, 300-301

271-72

Mercury,

ix, x, 10,

MESUR.

See

25

Mars Environmental SURvey

Meteorite (ALH84001), 207

ments, 53 of,

MFEX.

72

Massachusetts Institute of Technology (MIT) Artificial Intelligence

Laboratory, 61-62

"Micro-Lunar- Rover Challenge" (Wilcox), 62 microrovers, 62-63

See Microrover Flight

Microrover Flight Experiment (MFEX), 107-23

Alpha Proton X-ray Spectrometer (APXS), 95,

battery power, 110

bends 55, 60, 61

Pathfinder Planetary Rover Navigation Testbed,

on

(tight)

required of cables, 115-16

"cable harness," 114

Central Processing Unit (CPU),

vehicle, 53

"Robby" (megarovers),

54,

63-64

contact switches, 121

"designing on the path of least regret," 110

Robbyjr.,57

"Donna's Rule," 108

robotic arm, 54

emblem

rocker-bogie rovers, 58

"flex-cable" technology, 115-18

"Rocky," 58

height problems, 118-19

Rocky

"igloo tunnel," 115, 116

3,

1 1

Cognizant Engineers, 107-8

52, 57, 58, 63

source,

93,

176-77

nineteen-inch rack, electronics, 53

"One-Hundred-Meter Milestone,"

Experiment

"Micro-Lunar- Rover Challenge" (Wilcox), 62

mobility optimization, 20, 21, 24, 57

power

Institute, 176

McDonnell Douglas, 220

Central Processing Unit (CPU), 53

legacy

253

testing, 252,

"behavior control," 62

62, 63

of,

153-54

problem," 110, 171

Rover Expo, 60

"killing the

Semi-Autonomous Navigation (SAN), time-

micro-D connectors, not matching to

intensiveness of, 60-61

cables,

1 1

flex-

328

Index

Microrover Flight Experiment

momentum of team,

Moons,

(cont.)

ix

Moore, Hank

107

Pathfinder lander, taking personnel from, 121—

"Cussedness of Inanimate Objects, The," 257-59 Exceptional Service Medal, 309

22

on Mars

Pioneer Circuits, 116, 117

living

power concerns, 110-1

Rover Control Workstation (RCW), 204-5

Radioisotope Heater Units (RHUs), 112, 133

testing,

rocker-bogie height problems, 118-20

centrifuge testing, 160, 161

"Rover Significant Events," 108

Imager

science payload people influence attempts, 108

living

solar array power, 110-11

stress

Mars Pathfinder (IMP),

138, 139, 142

software, 187-88, 189, 191, 192, 194

MRSR.

18-20

RNET 9600 radio modem,

See

170-71

Mars Rover Sample Return

MSM (Mars Science Microrover), 66-67, 68-71,

on team, 108-9

team member dynamics, 109-10

72-

80,81

management, 108-9

technical problems,

for

on Mars time, 297

Motorola

"standup" concepts, 120-21, 190 1

257-59

Morrison, Jack

requirements, changing, 108

stowing the rover,

time, 289

Muirhead, Brian, 301, 306

thermal control, 111-13

Warm Electronics Box (WEB),

"Name

111-13, 115

See also Centrifuge testing;

Imager Noise

for

Software; Tethered

vs.

Miller,

101, 102-3, 105

cost condition of, 86, 195

Exceptional Achievement Medal, 309

210-11

Dave

failures, string of,

Mars Environmental SURvey (MESUR)

"faster, better,

Mars Rover Sample Return (MRSR),

96

cheaper,"

xiii-xiv, 12, 92, 100,

122,305-6,311

Pathfinder, 66, 68, 69, 71, 73, 79

team

at JPL,

308-10

Code R funding, 80-81, 100, Code S funding, 100, 103

untethered rover

Microrovers, 62-63 error,

ceremony honoring major achievements

Rovers (Marie Curie, Soft-

ware Development Model, Sojourner);

Midnight

(NASA)

Communications;

Mars Pathfinder (IMP); Lander;

filtering;

the Rover" contest, 132-33

National Aeronautics and Space Administration

wiring of rover, 114-16

Jet Propulsion

61, 62

building, 89, 90

Laboratory (JPL) and,

9, 10, 11,

12

Mishkin, Andrew, 1-2

Mars Rover Sample Return (MRSR),

Mission phases, 185-87

Microrover Flight Experiment (MFEX), 122

"Mission success panorama," 254

planetary protection policy

Mobility optimization of rovers, 13-24

rover missions (new), 305

of,

52, 58

166

"Blue Rover," 15-17

rover review, 152

bogie suspension, 17-18, 19-20, 20-21

workforce reductions mandated for future

bumps

vs. steps,

23-24, 57-58

rover missions, 305 See also Jet Propulsion Laboratory (JPL);

control issues, 16

"dome" wheels, 22

Launch

extraterrestrial vehicles (planetary rovers),

Software Development Model, Sojourner)

to Mars; Rovers (Marie Curie,

Naval Ordnance Test Station (China Lake), 25

14-15

"Mars Rover Sample Return" (MRSR),

20, 21,

Neptune,

ix, x, x,

10

Nguyen, Tarn

24

pantograph, 21-24

centrifuge testing, 159-60, 161

repair issues, 24

software, 185, 187, 190-91, 192, 194

Surveyor Lunar Roving Vehicle (SLRV),

15,

issues, 24, 54, 59, 65

See also

Computer- Aided Remote Driving

Noise

filtering,

175-84

176-84

Copper Room, 182

(CARD)

"MODEM_POWER_B" command,

26

Alpha Proton X-ray Spectrometer (APXS),

weight

167

"good enough" spectrum, 183

1

Index

high-frequency noise components as cause of

329

Pathfinder Planetary Rover Navigation Testbed, 52,

noise, 184

power

57, 58, 63

supplies

and cables

as source

of noise,

from software, 188-89

Personality

Petals of lander, 147, 248, 252

180-81, 184 spectra, noisy, 177-82

Phobos,

xi

Pioneer Circuits, 116, 117

Ohm,

Planet alignments for launch, 217

Timothy, 67, 70

Olympus Mons,

Planetary Society, 132, 133

xi

"One-Hundred-Meter Milestone,"

55, 60, 61

Pluto,

x

ix, x,

Operational scenario for rover, 197-99

Poem on Martian Local

Operations Readiness Tests (ORTs), 237, 239-43,

Polar ice caps of Mars, xi

244, 245-55, 257-60

Solar

Time

(Smith), 286-87

Polymorphic Systems, 28

Power

Operations team, 232-43

constraint, 85, 86, 91-92

"Powered descent," 145-46

'Amboys, The," 243 analysis team, 234, 235

Press conferences

command sequences written by,

234, 235, 236-

37,238

landing on Mars (July living

on Mars

4, 1997),

279-80

time, 285, 286-87

coordination with lander team, 234

Primus, Howard, 29-3

crossover meeting, 235

Public interest in mission, 202-4, 206, 288-89, 304

Data Controller, 234, 235

Pyrotechnic firings to get Pathfinder to surface, 250

Deep Space Network (DSN), 236 downlink team, 234, 235

Radioisotope Heater Units (RHUs), 112, 133

end-of-sol images of rover, 236

Radio signals

Experiment Operations Working Group meeting,

235-36

field-testing,

Mars (weak), 226-27

radios as commercial products, 169-70, 195

240-43

tethered

health assessment of Rover, 234

See also

Operations Readiness Tests (ORTs), 237, 23943, 244, 245-55,

257-60

Ramps

vs.

untethered rover, 97-98, 105, 169

Communications; Noise

238-39

RCW See Rover Control Workstation

Rover Control Workstation (RCW), 234

"Red Team Review," 93-95

Rover Coordinator, 234, 235, 238, 260

Reeves, Glenn, 167, 168

Rover Driver, 234, 236

RHUs (Radioisotope

Sequence Planner, 234, 236

Rieder, Rudi, 176-77, 178-79

"sol" (Martian day), xi, 209-11, stress

filtering

for rover, 147-50, 249, 252-53, 255

"Rapid prototyping," 126

command sequences,

review of

cruisin' to

234-35

on team, 239

Heater Units),

"Robby" (megarovers),

54,

112, 133

63-64

Robotic reconnaissance vehicles for military,

30,

31-33,37

turn rate sensor "drift" problems, 240, 292

75-76, 89-90

uplink team, 234, 236

Robotics groups, JPL,

See also Testing

Robotic Technology Test Vehicle (RTTV) program,

48-50

O'Rourke, Fran, 203

ORTs

(Operations Readiness Tests), 237, 239-43, 244, 245-55, 257-60

planets,

ix,

Rocker-bogie rovers capabilities of, 94, 104

Mars Environmental SURvey (MESUR)

"Outburst of the Rover Drivers," 301-2

Outer

14, 15, 29, 39,

x

Pathfinder, 69

Mars Rover Sample Return (MRSR), 58 Microrover Flight Experiment (MFEX),

Pantograph, 21-24 Parkyn.Jim, 171, 172-73

Rock Garden, 294-95,

Pathfinder. See Jet Propulsion Laboratory (JPL);

Rock-naming, 289-90

1

18-20

299, 301

Lander; Launch to Mars; Mars Environ-

"Rocky" 58

mental SURvey (MESUR) Pathfinder;

Rocky

Rovers (Marie Curie, Software Develop-

"Rocky 4" (Mars Science Microrover), 66-67, 68-71,

ment Model,

Sojourner);

Team

building

3, 62,

63

72-80, 81

330

Index

Rover Control Workstation (RCW), 195-206

Cedar

Wood Elementary School and,

Channeled Scabland (Washington

lander impact, minimizing, 150-52

203-4, 206

State) for

Mars geology, 202-3

command sequences

"Name

the Rover" contest, 132-33

Radioisotope Heater Units (RHUs), 112, 133 "rapid prototyping," 126

for rover, 195, 197-98,

Software Development Model (SDM), 126-27,

199,201,202

158

Computer-Aided Remote Driving (CARD), 198

Spaceflight Operations Facility, 129

debugging, 201-2

System Integration Model (SIM) "Marie Curie,"

"driver's test" for Brian field-testing,

Go-For

2,

131-32, 133, 155, 156-60

Cooper, 204

204-5

wiring boards, 125 See also Centrifuge testing;

203

input device, Spaceball, 198-99

Imager

"Mars Yard," 204

rity;

for

Communications;

Mars Pathfinder (IMP); Job

secu-

Lander; Launch to Mars; Mars Envi-

Mission Support Area (MSA), 206

ronmental SURvey (MESUR) Pathfinder;

operational scenario for rover, 197-99

Micro rover Flight Experiment (MFEX);

operations team, 234

Noise

public awareness of mission, 202-4, 206

(RCW); Software; Team

computer

Silicon Graphics, Inc.

for, 196,

197

vs.

filtering;

Rover Control Workstation building; Tethered

untethered rover

292-94

Spaceball as input device, 198-99

Rover team

Space Flight Operations Center, 205-6

Rover

stereo images, potential problems with, 204-5

RTTV (Robotic Technology Test Vehicle) program,

"strawman"

command sequence, tests,

revolt,

lander team, 244-45

48-50

201

terrain, position of rover in, 198

thermal-vacuum

vs.

Ruoff, Carl

Computer-Aided Remote Driving (CARD),

199-202

Rover Coordinator, 234, 235, 238, 260 Rover Driver, 234, 236

mobility optimization of rovers, 14-15, 16

Rover missions (new), 304-12

Semi-Autonomous Navigation (SAN),

"faster, better,

See Semi- Autonomous Navigation

Mars Exploration Rover mission, 311-12

SAN.

San Martin, Miguel

3

1

Mars Surveyor Program, 307-8 Space Shuttle Columbia

disaster,

cruisin to Mars, 224, 227, 229, 230

job security, 213

312

success, victims of, 305-6, 307

"Sarcophagus" for Sojourner

2001 rover mission, 305, 306-8, 311

Saturn,

ix, x, 10,

airline travel,

team, 286-88

2003 rover mission, 3 1

Scheduling

SDM (Software Development Model),

year-long mission, 305

"Self-righting" rovers, 67-68

Rover Navigation Imaging (movies), 273-74, 276, 277-78

difficulties for

126-27, 158

Semi-Autonomous Navigation (SAN), 38-51 artificial intelligence (AI),

Rovers (Marie Curie, Software Development

211-12

11

workforce reductions mandated by NASA, 305

camera mount

for,

40

44-45, 47

color graphics workstation, 49-50

Model, Sojourner), 124-33

dead reckoning

clock, 128-29

deployment, landing on Mars (July

4, 1997),

error,

43-44

Director's Discretionary

Fund (DDF), 40-41

goggles with liquid-crystal shutters, 49-50

274-77

80C85 microprocessor, 91-92,

126, 188

Humvee project,

48-50

maps of terrain, 42

electronic components, 125-26

local

Engineering Model electronics boards, 127-28,

machine vision software, 42

maps

129-31

flight electronics fabrication, 127-28,

Rover (FUR) "Sojourner,"

160, 161, 189

for "global" route of vehicle, 41-42

measurement uncertainty

female gender of rover, 132

Flight Unit

39, 40

cheaper," 305-6, 311

Mars Sample Return,

countdown

29,

30,31

129-30 132, 157,

error, 43

ninety-degree field of view, 47

path planner, 42, 47, 48

Remote Manipulator System for Space

Shuttle, 39

Index

Robotic Technology Test Vehicle (RTTV)

331

Software, 185-94

APXS Deployment Mechanism (ADM)

program, 48-50 sense-perceive-plan-act sequence of activities,

contingency sequence, 186, 187

41,48

shadows terrain

maps, 42

in

error messages, troubleshooting guide, 194

matching of maps,

hardware

42, 47, 48

tirne-intensiveness of, 60—61 visual tracking to

vs.,

191-92, 193-94

hazard detection system, 135. 136, 140-44, 186,

compensate

for

dead reckon-

ing error, 43-45, 47

294, 295

mission phases, 185-87

Sense-perceive-plan-act sequence of activities, 41,

48

personality from, 188-89 reliability of, 188

Separation connector, lander, 150-52

simplicity of, 188

Sequence Planner, 234, 236

stand-up

Shirley,

and,

191, 193

Donna

failure, 191,

192

testing, limited, 191-92, 193

communications, 169

Software Development Model (SDM), 126-27, 158

"Donna's Rule," 108

Sojourner, xin-xvi

Imager

for

Mars Pathfinder (IMP), 138

landing on Mars (July living

4, 1997),

Ares

267, 269, 279

Vallis (landing site), 2-3, 202,

Rover (FUR),

Flight Unit

on Mars time, 300-301

189

Mars Environmental SURvey (MESUR)

health checks, 227-31, 234, 259-60, 264

naming

Pathfinder, 80-81

132-33

of,

Mars Rover Sample Return (MRSR), 55-56, 62

public interest

Microrover Flight Experiment (MFEX), 107,

Rover Team, xv-xvi

in,

202-4, 206, 288-89, 304

stamp commemorating mission, 304

108, 122, 195

mobility optimization of rovers, 21-22

success

Rover Control Workstation (RCW),

uniqueness of team, 306

195, 196,

204-5

Development

ment Model, Sojourner)

105-6

"Sol" (Martian day),

xi,

209-11, 234-35

Solar system and Mars, ix-xii,

270, 297

x

Silicon Graphics, Inc., 49, 196, 197

Spaceball as input device, 198-99

SIM. See System Integration Model

Space business and JPL,

"Single string" components, 92, 152

Space Shuttle Challenger

Sirota. Allen

Space Shuttle Columbia

centrifuge testing, 158. 160. 161-62, 163, 164

Spear,

2, 8,

306

disaster.

disaster,

on Mars

time, 301

living

job security, 211

noise filtering, 179, 183, 194

launch to Mars, 222

rover missions (new), 307

on Mars

time, 285

tethered

Microrover Flight Experiment (MFEX), 114, 116-18, 122

vs.

untethered rover, 97-98, 99, 100.

103, 106

Sputnik,

noise filtering, 175-76, 177, 178, 179, 182, 183-84

9.

25

Stand-up of rover

operations team, 239, 241

design concepts, 120-21, 190

Rover Control Workstation (RCW), 200, 202

failure, 191, 192

landing on Mars (July

Sixth

Mars Science Working Group, 71-74

SLIM

(Surface Lander Investigation of Mars), 7172,

Smith, Peter, 135, 286-87

testing,

4,

1997), 274-75. 277

250

"Stereo waypoint designation" system, 32-37

73-74

SLRV (Surveyor Lunar Roving Vehicle),

11,96

312

Tony

cruisin to Mars, 227, 228

living

Mars Environ-

Rovers (Marie Curie, Software Develop-

building, 86-89, 90, 93

mode" of rover,

to Mars; Mars;

mental SURvey (MESUR) Pathfinder;

tethered vs. untethered rover, 97, 98, 100, 1034,

304

Launch

der;

Model. Sojourner), 132

"Silent

of,

See also Jet Propulsion Laboratory (JPL); Lan-

rovers (Marie Curie, Software

team

299

132, 157, 160. 161,

15,

26

Stewart- Warner, 18-19 Stone, Ed, 308

332

Index

Stone,

Henry

flight

landing on Mars (July living

on Mars

1997), 269, 275,

4,

276-77

microrover, 86

launch vehicles, 85, 86 length of mission, 93, 256-57

time, 282, 285

Microrover Flight Experiment (MFEX), 109,

Mars Observer disappearance and, 96 mass

121, 122

operations team, 238

constraints, 85, 86

mission

Rover Control Workstation (RCW),

196, 200,

power

206 rovers (Marie Curie, Software

Development

Model, Sojourner), 124-25,

126, 128,

129-

of,

93

Mobility-Thermal-Mechanical Subsystem, 87-88 constraint, 85, 86, 91-92

Power Subsystem,

87,

94

"Red Team Review," 93-95 science integration team, 95

31

components,

software, 185, 187, 189

"single string"

team building, 87

Telecommunications Subsystem, 87

tethered Stress

volume

259

testing,

vs.

untethered rover, 98, 99

constraints, 85, 86

See also Centrifuge testing;

on team

Imager

for

Lander; Launch; Mars Environmental

rity;

Microrover Flight Experiment (MFEX), 108-9

SURvey (MESUR)

operations team, 239

Flight

Pathfinder; Microrover

Experiment (MFEX); Noise

filtering;

Rover Control Workstation (RCW); Rovers

171-72,272

(Marie Curie, Software Development

"Structured light," 135, 139-40

Model, Sojourner); Software; Tethered

Success dedication of team

to,

Team member

Sojourner, 304

Temperature

victims

Sukamto,

of,

dynamics,

MFEX,

van Nieuwstadt, Lin

tethered

vs.

untethered rover, 98

visual noise and, 138, 139

x, x, xi

Sun sensor problems, 223-25

Temperature variances of Mars,

Surface Lander Investigation of Mars (SLIM), 71-

Testing, 244-60

72,

109-10

radios impacted by, 171-72, 269-70, 278-79

305-6, 307

Lin. See

vs.

untethered rover

213

requirements of mission, met, 286

Sun,

Communications;

Mars Pathfinder (IMP); Job secu-

job security, 208-9

Stride, Scot,

92, 152

xi

Alpha Proton X-ray Spectrometer (APXS),

73-74

"Surface Operations

Mode"

test,

258-59

166-67

Surveyor Lunar Roving Vehicle (SLRV),

15,

26

Surveyors, 10, 15

System Integration Model (SIM) "Marie Curie,"

APXS Deployment Mechanism (ADM), "common enemy" paradigm, 244-45

249

confidence, instilling in team, 259

contingency scenarios, 255

131-32, 133, 155, 156-60

"Cussedness of Inanimate Objects, The"

Tank Automotive Command (TACOM),

33, 37,

48-49

Landing (EDL), 246-49

Imager for Mais Pathfinder (IMP), 248-49, 250

Tarsala.Jan

"insurance panorama," 248

communications, 172-74 landing on Mars (July

4, 1997),

268, 269-70,

vs.

rover team, 244-45

Operations Readiness Tests (ORTs), 237, 239-

Team building, 85-96 (rate

lander

"mission success panorama," 254

272, 278-79

"burn rate"

(Moore), 257-59 Entry, Descent,

43, 244, 245-55,

of spending), 86

Cognizant Engineers,

87, 90, 93-94,

107-8

257-60

petal deployment, 248, 252

254-55

Control and Navigation Subsystem, 87

plant (potted)

"core team" meetings, 90

pyrotechnic firings to get Pathfinder to surface,

cost condition of

NASA,

86, 195

cheaper,"

122,305-6,311

250

ramp deployment,

design questions, 90-93 "faster, better,

in,

xiii-xiv, 12, 92, 100,

249, 252-53, 255

Rover Coordinator, 260 rover vs. lander team, 244-45

1

Index

333

landing on Mars (July

self-diagnostic of Sojourner, 260

Sojourner health check, 259-60 sol

1

Viking, 10, 146

Volume

stand-up of rover, 250 See also Centrifuge testing; Operations vs.

267, 268, 271,

Venus, ix,x, 10-11, 12

scenario, 247-50

Tethered

4, 1997),

272

software, limited, 191-92, 193

team

untethered rover, 97-106

constraints, 85, 86

von Karman, Theodore, 8-9 Voyagers

1

and

2, 7, 10,

281

Carnegie-Mellon University (CMU), 105 Central Processing Unit (CPU) location, 98, 99

conceptual design of rover, 97

Wakeup of rover, 160-61, 163 Wakeup songs for rover, 290

Dante, 105

Waldron, Ken, 20

four-wheeled rover, 101

Walking vehicles, 59-60

Instrument Deployment Mechanism IDM),

Wallace, Matt

landing on Mars (July

104, 105

radio links, 97-98, 105, 169

living

4, 1997), 264,

266

282, 284, 293, 297, 302

'separating out" of signals, 99

noise filtering, 178, 179, 180, 181, 182

temperature control of rover, 98

operations team, 238

time lost from controversy, 99

rovers (Marie Curie, Software

Tetrahedron (four-sided pyramid) lander, 146-47

Thermal

control,

MFEX,

Thermal-vacuum

tests,

111-13

testing,

Weight

landing on Mars (July

on Mars time,

4,

277

1997),

293, 297

Microrover Flight Experiment (MFEX), 116, 153, 154

Weisbin, Charles, 87

Welch, Rick landing on Mars (July

on Mars time,

4, 1997),

operations team, 238-39

Thompson Loop, 296-98

rover missions (new), 306

234-35. See also Living

Wheel

276

291, 297

operations team, 241, 242

turn rate sensors problems, 240, 292

111-13, 115

issues, 24, 54, 59, 65

living

Time on Mars, xi, 209-1 1, on Mars time

111-13, 115

217, 218-19

WEB (Warm Electronics Box),

167, 168

cruisin to Mars, 228, 229

living

256-57

Weather impact on launch,

communications,

Development

Model, Sojourner), 133

Warm Electronics Box (WEB),

199-202

Thompson. Art

Tilt

on Mars time,

abrasion experiment, 291

WTiittaker, William "Red," 59-60, 105

Wilcox, Brian

"Tooth," 62, 73

centrifuge testing, 155

Truth, Sojourner, 133

Computer- Aided Remote Driving (CARD),

2001 rover mission, 305, 306-8, 311

2003 rover mission,

3

Imager for Mars Pathfinder (IMP),

Socialist Republics, 9

University of Chicago, 176

Pathfinder, 66, 67, 68, 77 53, 54, 61,

62

on Mars

time, 285-86, 297

operations team, 234, 236

Uranus, ix-x,

265

Mars Rover Sample Return (MRSR),

Uplink team

revolt of,

4, 1997),

Mars Environmental SURvey (MESUR)

University of Arizona, 273, 276

living

134. 135,

136, 137, 138-40, 142

1

landing on Mars (July

Union of Soviet

25,

26-29,29-30,31-36

Tsou, Peter, 113

mobility optimization of rovers, 15-16, 20

Semi-Autonomous Navigation (SAN),

292-94 x,

43, 44-45, 46

team

10

building, 90

Wilcox, Howard, 25-26, 28, 29

Wilson, Bob, 101. 102. 103, 104-5

Valles Marineris, xi

van Nieuwstadt, Lin (Lin Sukamto)

communications,

170, 173

Wood, Gordon, 265 Wyle

Labs, 156-57, 158, 162

39. 40.

PHOTOGRAPHIC CREDITS

All

photographs in

this

book, unless otherwise noted, are courtesy

NASA /Jet

Propulsion Laboratory /California Institute of Technology. Images of the So-

journer rover, the names Sojourner®, Mars Rover®, and the spacecraft design are copyright

©

1996-97, California Institute of Technology, with

all

rights reserved,

and further reproduction prohibited.

the

The unofficial Sojourner rover patch design is courtesy Calvin Patton. The diagram of the solar system was created by the author. The positions of planets on July 4, 1997, were determined using SOAP (Satellite Orbit Analysis

Program) software developed by the Aerospace Corporation.

629.43 MISHKI Mishkin, Andrew. Sojourner an insider's view of the Mars Pathfinder :

MURRAY MBRARY 166 East 5300 South Murray,

c>^>

UT 84107

Andrew Mishkin

is

a senior systems engineer at

the Jet Propulsion Laboratory, where he has

w

&& ^

IP* if

coordinated the development of various robotic vehicles

and

their

fifteen years.

at

its

subsystems for more than

He joined

formation, eventually leading the rover

commanding the

operations team and

during the

the Sojourner rover team

its

rover

exploration of Mars. In 1997, he received

NASA Exceptional Achievement Medal and

was also selected

as

Made the Year" in

one of "The 35 People Who

the

December edition

Vanity Fair. Andrew is continuing

his

of

work as

the manager and designer of the Mission

Operations System that will control the next generation of Mars rovers, scheduled to land

in

early 2004.

Jacket design by Steven Ferlauto

Jacket photo of Sojouner courtesy of NASA/JPlVCalteeh Jacket image of Mars courtesy of NASA/USGS

Visit

our website at

www.penguin.com

A BERKLEY BOOK by The Berkley Publishing A division of Penguin Group (U

iblished

375

Hudson

Street

New York. New York 12/03

10014

roup inc.

About

a

minute be1H^K&i4f*ii#iti4Aii"@MpPr.

countdown. We brighter.

And

all

stared at the Delta.

Then the base

I

heard

someone

reciting a

of the rocket suddenly got

the rocket was moving.

Everyone cheered and applauded.

There was no sound yet from the sky so bright that

wondered

if I'd

I

see

thought all

rising rocket.

It

was too

should look away but

I

right after this

was

far

away.

A small

didn't.

It

climbed

part of

into the

my mind

over.

Evervone cheered again.

The sound

finally

reached us across the water.

A

kind of staccato

roar,

the voice of

pure power. The flame of the engines was rapidly becoming a bright red sky.

Pathfinder

was

traveling

was the spacecraft was

The champagne their faces

and

tiny

bottles

didn't

more than

were coming

even

a mile

now hundreds

realize

it.

out.

and a

half per second.

of miles away.

Some

I

star in

The spot

lost sight of

the

that

it.

people had tears streaming down

With Pathfinder on

its

way

to Mars,

many

people's jobs were complete.

Mine was

just starting.

Andrew Mishkin has been a key player in NASAs robotic program for the past fifteen years. He joined the Sojourner rover team at its formation, eventually leading the rover operations team and In

1997 he received the

NASA

commanding

the rover during

its

Exceptional Achievement Medal.

INCLUDES PHOTOGRAPHS

exploration of Mars.