Darpa's desert duel

BARSTOW, Calif.-If machines will one day take over the world, as science fictions writers predict, today may be remembered as their inauspicious start.

Very inauspicious.

The Defense Advanced Research Projects Agency's first self-directed vehicle race kicked off to a shaky start this morning, with only two vehicles out of the 13 starters still in serious contention , a mere five hours after the race's start.

The race began at 6:30 a.m. at the Slash X Caf', a tavern eight miles south of Barstow, Calif. One by one, 13 self-directing, or autonomous, vehicles set off a journey to find their way across 140 miles of Mojave desert. Two vehicles, from Team TerraHawk and the self-guided motorcycle from the Blue Team, scratched at the final moment, citing technical difficulties.

What makes this race the first of its kind is the fact that once the vehicles begin their journey, they must find their own way to finish line. No intervention from the team is allowed (although each vehicle is followed closely by a control vehicle with a' DARPA official who determines when the vehicle is out of commission or could cause harm. Each control vehicle has a kill switch for the autonomous vehicle it is following).

The two vehicles still in motion are a modified OshKosh commercial off-road military truck overseen by Team TerraMax, and a pickup truck from the Golem Group. Both were dark horses of the race, being placed late in the starting line-up.

What lies ahead of these machines is a jagged terrain with steep slopes, dry lakebeds, water, mud, cattle guards and, of course, lots of sand and rocks.

In addition to those two vehicles still running, three other vehicles on the course have not been disqualified, although they are in "thinking mode." In other words, they are not moving but are computing which way to move forward. They vehicles are from Team Caltech, SciAutonics I and Digital Auto Drive. The vehicles must finish the course within 10 hours. Too much contemplation will insure defeat.

Another three possible contenders are three vehicles that failed the initial start, but' were given a second launch at the end of the race. Those vehicles were from the Palos Verdes High School Road Warriors, Team CajunBot and the Team Ensco.

The vehicle most favored to win, a well-equipped Hummer from Carnegie-Mellon's Red Team, died a smoking death seven miles from the starting line. This team appeared to put the largest investment into its vehicle, an estimated $3 million worth of parts and labor. Out of the chute, which consisted of a pair of concrete barriers, the vehicle drove itself with impressive ease. It smoothly made its way down the dirt road, around a series of curves and out of site with little difficulty.

As the course got more treacherous, however, the Red Team's vehicle experienced a series of problems that eventually disabled it. First, it knocked out its front fender, losing its sensors and leaving only less-accurate global positioning satellite coordinate to navigate with, said Jay Gowdy, the technical adviser for the team. As a result, the Hummer soon wedged itself on top of a rock. Trying to free itself, it tore off the rubber off its tires. The rubber landed near the engine and caught fire, Gowdy said.

"No one said this was going to be easy," said Philip Koon, another member of the Red Team, which made it seven miles into the course.

Team CajunBot's vehicle, a modified all terrain vehicle ran, into a concrete barriers and tried to climb over a bush, jamming itself against the wall.

Team ENSCO's six-wheeled vehicle speedily jumped out of the starting gate and took off down the straightaway. It soon jumped up into bushes along side the road and, as it attempted to get back on the road, flipped over.

The vehicle may have been engineered to go a bit faster than was necessary, admitted Jacinda Clemenzi, an ENSCO team member, as the vehicle was being taken off the course by a tow-truck.

The race, called the Darpa 2003 Grand Challenge, was commissioned by Congress. The Defense Department staged the event in hopes that it would spur more interest in developing technologies for self-guided vehicles, which would help move large convoys of supplies through dangerous terrain.

Although the vehicles at this race are far from being able to think for themselves, they were all designed to do one thing intelligently on their own-find their way from one point to another. Such a job requires a lot of sensors and computational power.

"Today is a good day for the tow-truck drivers," one spectator said.

Click to link to an interactive video presentation from washingtonpost.com's TechNews.com.

(Posted March 13 and updated 7:56 a.m. March 15)

About the Author

Joab Jackson is the senior technology editor for Government Computer News.

inside gcn

  • artificial intelligence (vs148/Shutterstock.com)

    Government leans into machine learning

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group