Select your localized edition:

Close ×

More Ways to Connect

Discover one of our 28 local entrepreneurial communities »

Be the first to know as we launch in new countries and markets around the globe.

Interested in bringing MIT Technology Review to your local market?

MIT Technology ReviewMIT Technology Review - logo


Unsupported browser: Your browser does not meet modern web standards. See how it scores »

{ action.text }

In the Beginning

The philosophy underlying “faster, cheaper, better” is “less is more.” In space flight, mass equals money-lots of it-and complexity means risk. To make, say, rovers affordable and reliable, create them to be as lightweight and simple as possible-25 pounds in Sojourner’s case, with a feeble but sturdy 8-bit, 1970s-level processing unit for a brain. As obvious as this strategy seems now, through the late 1980s NASA’s Jet Propulsion Laboratory (JPL) focused on building something completely different: a truly formidable machine in size, range, and computational and operational ability-and cost. This Mars Rover Sampler Return (MRSR) vehicle would measure up to 8 feet long and weigh half a ton. It would be built to cruise for a year and a half in Mars’s punishing cold, across 700 miles of its varied, rugged terrains, gathering samples and blazing a path for human missions. It would cost up to $10 billion.

“Then fiscal reality hit,” recalls David Lavery, manager of NASA’s telerobotics research program. “We realized it just wasn’t going to happen.”

Luckily, unbeknownst to the NASA brass, an alternative waited in the back laboratories at JPL-a neglected experiment in robot automation and simplification. Even as JPL as an institution was chasing the MRSR dream to its dead end, a small group of upstart rover and another group of artificial-intelligence renegades, both at JPL, had been quietly seeking solutions of their own.

The effort began in 1988, when Howard Eisen, now chief mobility engineer for JPL’s rovers, left his graduate studies at MIT to work at the lab. He brought with him a particularly apt thesis project: building a one-eighth-scale model of the mighty MRSR. He found that the model, guided by an electric tether, performed much better than expected. Indeed, its five-inch wheels could climb over objects up to eight inches high. So perhaps the jumbo MRSR was unnecessary, he thought; could “a much smaller platform” negotiate Mars’s rocky surface?

Eisen and his JPL colleagues set out to build such a platform-working on their own in the garage of engineer (and former hot-rodder) Don Bickler. After several trials Bickler invented a six-wheeled chassis that could maintain even weight and traction on all its wheels. The engineers called it a “rocker/bogey” after its two key mechanical elements and with all due honor named the subsequent rover prototypes Rocky. “Everyone was joking over whether we’d have as many sequels as the Rocky movies,” Eisen recalls. (They would.)

Midway through building the first Rocky, the rover renegades obtained lab space at JPL and gained new collaborators: a team of artificial-intelligence (AI) designers. Inspired by the “subsumption architecture” approach of MIT’s AI pioneer Rodney Brooks-designing robots to operate using a hierarchy of simple reactions to stimuli-David Miller, a recent arrival at JPL’s AI section, hired Brooks’s student Colin Angle for a summer. At MIT Angle had created the trailblazing Genghis robot, which despite its low brainpower could autonomously perform a fairly complex function-gathering up all the coffee cups in the office. (That machine is now in the National Air and Space Museum.) At JPL, Angle built a similar robot named “Tooth” using a model-car chassis, for less than $500 in parts and $5,000 in labor. “My only constraint,” Angle recalls, “was I couldn’t spend more than $50 for each part, so it could all come out of petty cash.”

Miller and his teammates saw the potential of the mechanical Rocky that Bickler’s group had developed. Marrying Tooth’s electronic “brain” to Rocky 3’s body, they created the first autonomous rover that could operate outdoors, on actual dirt.

JPL managers were impressed, but still wedded to their big MRSR. Then Congressional members howled at its cost and that project was dead. NASA had meanwhile found funding for a single small Pathfinder lander and cast about for something to carry on it. “I said, We happen to have this one rover,’ ” recalls Miller. At last the renegades had an actual mission to work on.

But not for long. As so often happens when innovations are channeled into the mainstream, the innovators were left out in the cold. Miller and his teammates lost control of the Rocky project, and all but one of them left JPL for the private sector. Rover budgets increased, and so did timelines. Miller’s group took just one-and-a-half years to progress from Tooth to the fourth-generation Rocky, but JPL then took five more years to advance to Sojourner, the sixth in the Rocky line.

JPL’s Brian Wilcox, who took the helm of the microrover project after Miller, argues that this was a natural response to the challenges of making technologies “reliable enough” to work in the harsh Martian environment. Indeed, novel as it seemed to TV viewers, Pathfinder was a rather conservative mission; JPL reached back to tried-and-true command technology rather than attempting fully autonomous operation. Better safe than venturesome when the whole world is watching.

0 comments about this story. Start the discussion »

Tagged: Computing

Reprints and Permissions | Send feedback to the editor

From the Archives


Introducing MIT Technology Review Insider.

Already a Magazine subscriber?

You're automatically an Insider. It's easy to activate or upgrade your account.

Activate Your Account

Become an Insider

It's the new way to subscribe. Get even more of the tech news, research, and discoveries you crave.

Sign Up

Learn More

Find out why MIT Technology Review Insider is for you and explore your options.

Show Me