No Drivers Wanted in Race for $1 Million

Think “Mad Max” meets Jules Verne. Or “BattleBots” hits “Cannonball Run.”

Think winning $1 million for racing a robocar.

That will be the Pentagon’s unlikely pitch to more than 200 potential participants Saturday in Los Angeles at the announcement of a public competition to build and race unmanned ground vehicles from Los Angeles to Las Vegas in March 2004.

The rules are simple. “No humans or other biological entities” allowed onboard. No radio or remote controls. No attacking other vehicles. And please, no flamethrowers or other devices that “clear a path by setting everything in its way on fire.” Pretty much anything else goes.

The race, called the Grand Challenge, is the brainchild of the Defense Advanced Research Projects Agency, or Darpa, the $2-billion whiz-bang shop at the Pentagon that helped create the Internet, Stealth aircraft, “smart” bombs and the pilotless Predator plane.

The goal this time is to meet a congressional mandate, set in 2000, that at least one in three future army battle systems be unmanned. Despite huge advances in civilian and military robotics in recent years, the necessary sensors, software and other technology for real robot-assisted warfare doesn’t yet exist.

The Pentagon hopes garage tinkerers, junkyard warriors, off-road enthusiasts, robotics fanatics and anyone else will come to the rescue by building a fully autonomous ground vehicle -- any size, any shape -- that can traverse 300 or so miles of paved road, deserts and mountains in less than 10 hours. That would require an average speed of at least 30 mph.

“We look at this like [Charles] Lindbergh crossing the Atlantic,” said Air Force Col. Jose Negron, who is running the Darpa race, referring to the $25,000 prize that spurred a shy mail pilot to make the first nonstop solo flight from New York to Paris in 1927.

“It’s mind-boggling,” Negron added. “This will be a spectacular event if someone builds an autonomous vehicle that can travel 300 yards, no less 300 miles. That will be inspiring. It’s never been done.”

The task is far tougher than building a ground version of the Predator, the missile-firing drone aircraft that earned public notice for missions over Afghanistan last year.

“There’s nothing to bump into up there except another aircraft,” said Negron, who was a B-52 navigator and thus should know. Plus, the unmanned Predator is electronically “tethered” to a technician somewhere on the ground who flies it by remote control.

The ground vehicle must be independent -- that is, all computing and other equipment must be self-contained and onboard. It can use satellite-based navigation systems, but no other external communication. The vehicle even must control its own refueling or recharging.

The precise course won’t be revealed until race day, so each roadster must be able to “read” the ground as it advances. It then must recognize and negotiate over or around steep hills, rocky arroyos, desert sand, craters, buildings and whatever artificial obstacles Negron and his team can dream up.

“Suppose it comes across a pond or a stream?” Negron asked. “How deep is the water? As a human being, I can see tracks across the way and figure I can cross it. But can an autonomous vehicle see that? Or will it know to go around it? And can it go around?”

Los Angeles was chosen as the starting point because of its car culture, its defense industry and because Darpa will hold its annual tech conference in Anaheim just before the race, now scheduled for March 13, 2004. Las Vegas was chosen as the finish line because, well, it’s Las Vegas.

Early discussions at Darpa have focused on everything from super-charged Humvees to huge air-powered balls.

Whatever the final configuration, the Pentagon needs unmanned vehicles to “reduce the number of soldiers placed in harm’s way and increase combat effectiveness,” according to a report issued last month by the National Research Council, the operating arm of the National Academy of Sciences and the National Academy of Engineering.

“Unmanned ground vehicles have the potential to revolutionize the capabilities of Army forces on the battlefield,” the report adds. It cites their potential use as scouts and searchers, as “donkey” units to move supplies and equipment, and ultimately as “a fully autonomous combat vehicle.”

To be sure, other groups have tested and raced robotic vehicles.

Carnegie Mellon University built a self-steering one that drove nearly all the way from Washington, D.C., to San Diego in 1995, using cameras to keep the vehicle on the highway and away from other cars.

However, researchers controlled the brake and the accelerator, which won’t be allowed in next year’s race.

Autonomous vehicles also have covered rugged terrain, though at low speeds. Carnegie Mellon’s Nomad, built for NASA, was able to scour a moraine in Antarctica three years ago to discover meteorites.

But the Darpa race is so demanding and requires such high speeds that several robotics experts said they doubt anyone will be able to clinch the $1-million prize for making the first robo road trip.

“This is a quantum leap in terms of difficulty compared to other competitions,” said Ronald C. Arkin, director of the Mobile Robot Laboratory at Georgia Institute of Technology in Atlanta.

Darpa plans to run the race again within two years if no one wins the first go-round. Chuck Thorpe, director of the Robotics Institute at Carnegie Mellon in Pittsburgh, said it may take even longer for a winner to emerge.

“I wish I could tell you it would be soon, because that would be exciting for the field,” Thorpe said. “But I don’t have confidence it will be soon. It’s a very tough challenge.”

The robo-race is in a tradition of grand challenges designed to inspire the public and push the frontiers of science.

In 1714, for example, the British Parliament offered a huge reward to anyone who could solve the key navigational problem of the day: how a ship at sea could determine its longitude. It took almost 60 years for a self-taught clockmaker from Yorkshire to collect the prize.

Paul MacCready, an aeronautical designer from California, won $100,000 in 1977 by designing a foot-powered, pedal-driven aircraft that flew around a figure-8 course. He helped win $200,000 more two years later by designing another pedal-powered plane that flew across the English Channel.

Contests also have a history in the world of robotics. Hundreds of experts and students devote themselves to building soccer-playing robots for an international competition called RoboCup, run annually since 1997. The next tournament is this July in Padua, Italy.

RoboCup organizers say their long-term goal is to field a team of robot soccer players by 2050 that can beat a top human team. But by setting such an ambitious and engaging goal, organizers also hope to push researchers to tackle technical hurdles in artificial intelligence, vision systems, multi-robot systems and the like.

For now, it’s not clear who will take up the Darpa challenge. More than 200 people have indicated plans to attend the competitors’ conference at 9 a.m. Saturday at the Petersen Automotive Museum at 6060 Wilshire Blvd. to hear details of the race and to meet potential collaborators.

Only a handful of actual entries are expected to result. But a forum on Darpa’s Web site https://www.darpa.mil/grandchallenge/overview.htm for the race carries inquiries from small companies, amateur inventors and science fiction fans looking for colleagues and support.

“I’m a robot-geek in Dallas/Fort Worth, seeking to form a team,” wrote one, posting a resume that includes gunsmithing and building robots that play paint-ball. “On the down side, my finances are limited [$100/month].”

That’s a problem. There is no charge to enter the race, but it could easily cost more than the $1-million prize to build the winning machine.

Whether defense contractors, Detroit automakers or other corporate sponsors will step up remains to be seen.

“What if they enter and a small research company or a team of hobbyists beat them?” asked Raffaello D’Andrea, associate professor of engineering at Cornell University. “I don’t think they have a lot to win, and they have a lot to lose.”

Christopher Wagers, president of VIDEOptions, in Lake Oswego, Ore., said he has six people working full time on a race vehicle and expects to commit $300,000 to the project. They are using a chassis from a Volkswagen Beetle, building a body from scratch, and outfitting it with a radar vision system.

“We want to show people that it can be done, that you can use your imagination, that you can do almost anything,” Wagers said.

Phineas Fogg couldn’t have put it better himself.