Robot race more than ordinary challenge

Friday, March 12, 2004 | 10:47 a.m.

As might be said in Vegas sports betting parlors, may the best unmanned autonomous vehicle win.

If the "headless Hummer" or one of several other robot vehicles cross the Mojave Desert and arrive in Primm within 10 hours of their departure from Barstow, Calif., on Saturday, the team that built the winning vehicle will win $1 million.

The Defense Advanced Research Projects Agency, or DARPA, the Pentagon's few-holds-barred research and development arm, is sponsoring the race, dubbed the Grand Challenge.

As for the contestants, they're a motley bunch of garage tinkerers, off-road enthusiasts, high school students, physicists and programmers who hope their microprocessor-jammed jalopies usher in the next generation of military combat vehicles.

"It's a marriage of the geeks and the greaseballs," said Sal Fish, a longtime desert off-road race promoter and the lead designer of the course. "If they go even two miles, I'll be in awe."

Twenty teams tested their mettle in early week qualifying runs for the so-called Grand Challenge, part of Pentagon efforts to have one-third of all ground vehicles unmanned by 2015. Five teams dropped out before Monday's first "road" test.

By Thursday night, six vehicles had gotten through the road test that was a little more than a mile long, though the robots don't necessarily have to finish the trial course to qualify for the race.

Today DARPA is to announce the teams that made it into the race.

Two hours before the race begins Saturday, DARPA will give competitors a CD-ROM with Global Positioning System coordinates that chart the eastward course.

A DARPA vehicle will be assigned to each robot contestant, with a judge ready to hit a kill switch if it goes astray. Helicopters will also monitor the action.

DARPA officials are considering several possible routes along dirt roads and rough trails, ranging from 150 to 210 miles. Even the shortest course would require the robots to average 15 mph, a feat that has eluded major defense companies.

Several of the robots are capable of 65 mph but obstacles along the way will require them to go much slower as they rely on combinations of orientation devices ranging from GPS satellite positioning to digital compasses, ultrasonic scanners and gyroscopes.

The race ends with an 11-mile run through the dry lake bed outside Primm.

If no one finishes on time -- a likely outcome, many participants say -- DARPA will host another contest, probably in 2006.

In addition to military uses, technologies spawned by the Grand Challenge could eventually find form in such inventions as collision avoidance systems for cars or automated farm equipment.

"It's not only conceivable, but it's expected that commercial applications for this technology will emerge," said Darrell Davidson, executive director of the Association for Unmanned Vehicle Systems International, which includes a number of major defense companies.

The most feared competitor is Carnegie Mellon University's "Red Team." The team has used cash and in-kind contributions valued at more than $2 million from heavyweight sponsors including Intel Corp. and Boeing Co. to arm itself with super-fast computers and sophisticated mapping software.

Team leader William "Red" Whittaker used robots to explore Antarctica in 1993 and map a 3,500-foot corridor in an abandoned Pennsylvania mine last year.

Sandstorm, the team's red 1996 Humvee, was the first vehicle to dodge all the bricks, metal rods and parked cars necessary to complete the road test.

Other contenders include: Bob, a modified 1996 Chevrolet pickup put together by the California Institute of Technology; Cliff, an off-road vehicle from Virginia Polytechnic INstitute, and SciAutonics II, an Israeli-made dune buggy created by a team based in Thousand Oaks, Calif.

Participants say the key to success is how well sensing devices -- including digital cameras, radar and sonar -- feed data to the onboard computers. If they work properly, the sensors should be able to tell a boulder from a tumbleweed and determine whether a ravine is too deep to cross.

But coordinating sensors with navigation and steering systems has been a daunting challenge.

"Everyone thinks they have a solution to the problem of how to handle the flood of data," said Bill Zimmerly, who wrote software for 28 years and helped design a Kawasaki ATV entrant. "At these speeds, no one has been able to do it."

One of the intriguing aspects of the event is the military offering a prize outside its normal channels of research and development to solve a military, technological problem: how to fight wars in the future that put few people in harm's way, using robots.

The same thinking led to the Predator, Nellis Air Force Base's unmanned planes used in Iraq -- and to the U.S. Congress recommending that one of three Army vehicles on the ground be a robot by 2015.

And the race is being held in a desert because of "obvious reasons," said Jan Walker, speaking for DARPA.

There is precedent for government-funded contests to solve scientific problems of the day, including the British Parliament's 18th century "Longitude Act," with its 20,000 pound sterling prize to whomever found a way to better measure longitude at sea. And some now support funding a round-trip to Mars through a $20 billion prize.

But big-government models of research like the Manhattan Project and the Apollo program have held sway in contemporary U.S. history.

Walker called meeting the congressional goal "a technological issue that is very difficult and challenging. We wanted to make sure we were aware of all good ideas out there that would have application to the challenge."

Another person close to the race put it this way: The Defense Department gets hundreds of graduate students to do a ton of research at a fraction of the normal cost.

The fact that the race may not have a prize-winner has not stopped the world's media from tramping to the desert, with more than 300 covering the event. The result has been to put robots on a par with Tom Cruise in certain glossy magazines.

Anthony Levandowski, who leads a team of 17 under-25-year-olds at the University of California, Berkeley, said he didn't mind the double-time hours he and his teammates are putting into the race. He figures the media coverage exposed the world to his team's "out of the box" idea -- what he says is the world's first autonomous unmanned motorcycle.

Levandowski said he asked his fellow students from the California home of free speech and pacifism -- "Does anybody have any problem working on something that could used in a couple of years to kill people?

"There was a moment of silence. Then everybody said, 'I have no problem.' "

The Associated Press

contributed to this report.