A souped-up Chevy Tahoe sports utility vehicle with a mind of its own was declared the winner of a robot car race on Sunday after it traveled without help from humans for six hours and 60 miles around a California ghost town.
Nicknamed Boss, the vehicle from Carnegie Mellon University of Pittsburgh won a $2 million prize in the third such race sponsored by the U.S. Department of Defense, which wants robots eventually to drive military supply vehicles.
The entrants — including station wagons and a huge green military truck all decked out with flashing lights, warning sirens, spinning laser range finders and cameras — looked like mini-versions of the "monster" trucks that duel at arenas around the United States. Only six vehicles out of 11 finalists finished the course on an abandoned military base on Saturday.
The winners were determined overnight based on safety as well as speed. Stanford University, which won a 2005 race, came in second and Virginia Tech finished third.
"Yesterday (was) a very historic day," said Tony Tether, director of the defense department’s research agency, after handing out checks on Sunday as the winning cars, with humans at the wheel, took off for a victory spin.
The effort has brought together some of the top talents at U.S. universities and corporations to work on a big technology challenge. Science fans have followed the robots’ progress avidly, but the biggest benefits are in the future.
The U.S. military aims to put robots behind the wheel of supply vehicles — with a goal of making a third of its supply fleet robotic by 2015 — to keep soldiers out of danger. And automakers see intelligent cars helping people drive and eventually taking over the task altogether for better safety and comfort.
The race’s objective — for the vehicles to finish without a dent, following California traffic rules precisely, within six hours — was daunting. A minor fender bender, the worst accident on Saturday, drew a collective gasp from hundreds of fans drawn to the abandoned base about 80 miles northeast of Los Angeles.
Vehicles with hundreds of thousands of lines of computer code for brains appeared to have their own personalities. The Carnegie Mellon General Motors SUV rushed out of the starting gate, while Stanford’s Volkswagen, named Junior, was conservative.
"Boss is kind of like a soccer mom with some place to be — aggressive but safe," said Carnegie team member Bryan Salesky.
The Toyota Prius from the University of Pennsylvania and Lehigh University took corners slowly and then accelerated.
The vehicles hit top speeds of about 30 mph (48 kph).
"You see the steering wheel move, and there’s nobody there. It’s mind boggling," said stunt driver Tammie Baird, who drove the streets along with the cars to test their traffic skills.
Using a combination of satellite navigation, cameras, radar and lasers, computer-based artificial intelligence systems determine where the car is and where to go, then deliver directions to a system which drives the car, from steering to acceleration.
Universities hope their technology will help cut traffic accidents and improve driving.
The Defense Department’s research arm, the Defense Advanced Research Projects Agency, known as DARPA, sponsored dirt-road robot races in 2004 and 2005. The 2007 Urban Challenge was much harder, requiring complicated decisions from robots that in previous years simply had to find a course and stay on it.
The Defense Department gave a little help when needed to robots racing on Saturday. It regularly used remote pause systems to halt vehicles, avoiding traffic jams that could have damaged or confused the cars.
One team was allowed to dig its car out when it took an unanticipated left turn off a dirt road and got stuck, while at least one computer system was restarted during the race.
That did not lessen the cheers from the crowd or enthusiasm from organizers on Sunday.
"To end up today with you guys going into traffic, with not only other people driving, but other robots driving, is just absolutely fantastic," Tether said, to wild applause.