BARSTOW, Calif. -- A race between Steeler Jerome Bettis, Olympian Marion Jones, wheelchair athlete Rory Cooper and your 10-year-old niece would not be any more bizarre than the contest that will commence at daybreak today in the foothills outside this Mojave Desert town.

	John Beale, Post-Gazette Spencer Spiker, head vehicle team leader for Carnegie Mellon University's "Red Team," climbs out Sandstorm, their autonomous robotic ground vehicle that will compete in the DARPA Grand Challenge. The robot had just been unloaded at the starting point in the desert in Barstow, Calif. Click photo for larger image.

A total of 15 robotic vehicles, led by Carnegie Mellon University's Sandstorm, will line up for the beginning of the Grand Challenge race sponsored by the Pentagon's Defense Advanced Research Projects Agency, DARPA.

The finish line will be 150 to 200 miles away in Nevada, where a $1 million prize awaits the team whose robot finishes first and in less than 10 hours.

"I wouldn't bet the house on it," said Jay Gowdy, one of 70 members of CMU's Red Team who are assembled here, "but I think there's a chance we may win it this year."

Sandstorm, a heavily modified Humvee, is the consensus favorite, awarded the pole position by the agency, based on its performance in two qualifying runs on an obstacle course this week in Fontana, Calif. It is packed with high-powered computers, global positioning satellite antennas, radar, laser rangefinders and an advanced stereo vision system.

But other teams took different approaches and it remains to be seen which team made the right choices.

Oshkosh Truck is using a giant six-wheeled military truck that can plow over most obstacles.

Anthony Levandowski, a young inventor from Berkeley, is attempting to navigate the course with a self-balancing motorcycle.

Another group, Digital Auto Drive of Morgan Hill, Calif., is taking the simple route, adding a couple of cameras to a Toyota pickup and tying them into a laptop computer.

"There are at least three or four teams that have a good chance of finishing the course," said Anthony Tether, director of DARPA, the Pentagon's research and development arm.

The goal of the race is to spur the development of robotic vehicles that can autonomously deliver supplies or roam dangerous areas scouting for enemy troops.

Twenty-five teams were invited to the qualifying event for Grand Challenge this week. Twenty-one actually showed, 19 attempted to run the obstacle course and seven made it through the course at least once.

The robots must negotiate the course today without any human intervention. They will start the race sequentially, with about a five-minute interval between each robot.

The starting order was determined not by speed -- though, at 6 minutes 45 seconds, Sandstorm was the quickest on the 1.4-mile obstacle course -- but by which teams seemed most likely to be successful.

After Sandstorm, a robot built by a Southern California team, SciAutonics II, will start, followed by Team Caltech and Digital Auto Drive. Virginia Tech, which modified a four-wheel-drive, golf-cart-size utility vehicle, will start fifth.

Safety is one of the major concerns on race day and DARPA will have 80 Marines, 250 to 300 DARPA staff members and 150 officials from Score International, a sanctioning body for desert racing, on hand today to keep the course clear of human and non-human life.

Desert tortoises, a threatened species, are one concern. Twenty biologists will scour the course one hour before the race to get any tortoises out of harm's way.

Each robot also will be trailed by three people in a chase vehicle, who will be able to stop the robot if a safety emergency arises.

Depending on the number of stops, which also will be made to allow robots to pass, it's possible that some robots won't have 10 hours of daylight to complete the race. If so, the race will be stopped at sundown and resumed tomorrow.

"The robots don't care whether's it's day or night," Tether said. "It's the people we're worried about."

Yesterday afternoon, the teams assembled in 80-degree heat at the starting line.

CMU's Red Team unloaded Sandstorm and prepared a 54-foot-long trailer, which houses 14 high-powered computer workstations that will be used this morning to prepare a detailed route plan for Sandstorm. The exact route wasn't announced to the teams until about 3:20 a.m. today.

Using highly detailed, computerized maps, a software program will spit out a proposed route for the robot. The human route editors will then fine-tune the route, suggesting, for instance, that Sandstorm hug the left side of the road in some places.

In some rough sections, Sandstorm may have to slow to 10 mph, but in wide open areas, it can go 35 mph or better, said Chris Urmson, the team's navigation leader.

"Most things we can take at 22 mph and it really doesn't care," he said.

And that's fast enough to win.