Just a few years ago who would have imagined that vehicles on the ground could be driven without drivers. The recent Defense Advanced Research Projects Agency (DARPA) driverless vehicle race has proved that it is possible to build unmanned vehicles. The DARPA Grand Challenge was the first race of its kind in which autonomous ground vehicles used nothing but onboard sensors and navigation equipment controlled by computers and artificial intelligence to steer themselves along the desert course in less than 10 hours. And, unlike traditional vehicle races that include mostly well-paved straight roads and curves, this race included tunnels, mountain switchbacks, lakebeds and on- and off-road stretches. In fact, this year's race was crucial because none of the participants were successful last year, raising some doubts about the feasibility of autonomous vehicles for long-range missions. But following this year's success, the hopes are high.
There were 23 finalists in the second DARPA Grand Challenge race, out of which only five unmanned vehicles crossed the finish line after a 132-mile journey through the tough terrain of Nevada's Mojave Desert. As soon as the autonomous vehicle race concluded in Primm, Nev., DARPA's director, Anthony J. Tether, said, “These vehicles haven't just achieved world record, they have made history.”
The $2 million prize went to an autonomous diesel-powered Volkswagen Touareg named Stanley, designed and developed by a team from Stanford University. The Stanford team included students, staff and professors from computer science, mechanical engineering, Aero-Astro and optimization. Some members of the team were affiliated with corporate research labs in the Bay area. The overall team leader was Sebastian Thrun, director of Stanford Artificial Intelligence Lab and associate professor of computer science. Stanley covered 132 miles of Nevada desert in six hours and 53 minutes, for an average of just a little more than 19 miles per hour.
As described in the Stanford report (www.stanfordracing.com), Stanley exploited an array of sensors plus GPS to determine its position within two inches. Lasers mounted on the roof continuously scanned the ground in front of the vehicle looking for cattle gates, ditches, barbed-wire fences or disabled cars on the course.
While the lasers let Stanley see only four car lengths ahead, the autonomous Volkswagen Touareg exploited long-range radar to scan the horizon. Stereo-vision cameras, mounted inside the car to avoid dirt spattering, took high-resolution pictures of the course, while computer algorithms “learned” the terrain and mapped out an optimal driving surface. Stanley's brains — seven Pentium M laptop computers — were in the trunk. The machines were shock-mounted to survive the bumps on the road, and the networked system had multiple copies of every program. A thick spinal column of electrical wires connected the computers to an electronic brake, throttle and chain-driven steering column.
Second and third prize in the DARPA race went to a pair of Hummers from Carnegie Mellon University. The Red Team's Sandstorm was second and H1ghlander came in third.
DARPA has sponsored autonomous vehicle research for more than a decade. But, progress has been slow. By bringing researchers and innovators from around the country and the world together, the agency has given unmanned vehicles a shot in the arm. Although this represents a giant step forward for military and critical missions, commercial applications are not far behind. Like many other military innovations and technology advancements that have benefited the commercial, industrial and medical world, the unmanned vehicle advancements are also bound to accelerate the advancement of safety and vision-sensing technologies in the commercial cars sooner than we think. Meanwhile, such driverless vehicles are under development for U.K.'s Heathrow Airport (www.atsltd.co.k), where in a few years, passengers could be riding autonomous vehicles from car parks to terminals.