Dusted: No winners in DARPA's $1 million robotic race across the MojaveDesert

But there were plenty of new ideas for military vehicle developers

5 min read

18 March 2004--Nobody said it would be easy. In fact, pretty much everybodysaid it couldn't be done. When the U.S. Defense Advanced Research ProjectsAgency (better known as DARPA) announced just over a year ago that it was sponsoringa robotic race, complete with a US $1 million prize, experts were skeptical.Autonomous vehicles of various stripes existed, to be sure, but what kind ofmachine could master a punishing 320-kilometer, off-road course through the Mojavedesert in under 10 hours, without benefit of human control? "Everyone thoughtit was impossible," recalls Ümit Özgüner, a professor ofelectrical engineering at Ohio State University, in Columbus, and founding presidentof the IEEE Intelligent Transportation Systems Council.

Photo: Courtesy of DARPA

Carnegie Mellon's Sandstorm

a modified Humvee, roars outof the starting gate at the Darpa Grand Challenge on 13 March. None of the 15 robotic vehicles came close to completing the 225-kilometer cross-desert course from Barstow Calif. to Primm, Nev.--or to claiming the $1 million prize--but Sandstorm came closest, breaking down in mountainous terrain about 12 km out. DARPA plans to hold the event annually, to help advance R and D on military-related autonomous ground vehicles, through 2007.

And yet, in the chilly pre-dawn hours of 13 March, at a biker bar called the Slash X Café south of Barstow, Calif., there Özgüner was, along with hundreds of other engineers, robotics experts, computer scientists, students, garage monkeys, and others who had decided, despite the incredibly long odds, to take up DARPA's gauntlet. From an initial list of over a hundred applicants, DARPA had narrowed the field, based on technical reviews and some site inspections, to 25 finalists. After nearly a week of nerve-racking trials, held at the California Speedway, in Fontana, 15 vehicles managed to qualify for the Grand Challenge, which DARPA steadfastly refused to call a race, despite all appearances to the contrary.

About three hours before start time, each team received a CD-ROM containing some 2200 Global Positioning System coordinates outlining a 225-kilometer course that snaked through the Mojave Desert from the Slash X all the way to the border town of Primm, Nev. Mapping experts then got to work, wherever possible seeking out the straightest, most navigable path and tweaking the route to avoid known obstacles. In the months and weeks beforehand, a number of teams had surveyed probable routes through the 3900-km2 area where the Grand Challenge would run, using satellites, airplanes, and cars. One team was even rumored to have monitored DARPA radio communications while the organizers drove around the desert finalizing the route.

By start time, dozens of TV crews, photographers, and reporters, some from as far away as Germany and Japan, were jockeying for position around the start line. At about 6:30 am, the first vehicle, Sandstorm, a cardinal-red Humvee fielded by Carnegie Mellon University's Robotics Institute, in Pittsburgh, roared out of the gate, and subsequent vehicles got underway, or tried to, every five to ten minutes. A DARPA chase vehicle tailed each robot, and if a robot went off course or appeared in danger of hitting something, the chase vehicle had a wireless "e-stop" device that would immediately shut it down. Each robot could also be "paused" to let other vehicles pass.

The first part of the course traced a large loop around the Slash X, and from the packed bleachers, spectators could see much of the first several kilometers. Only a few vehicles made it that far, though. Two dropped out even before the event got under way, and a number barely made it out of the starting gate. The brakes on Virginia Polytechnic Institute's golfcart-size robot locked up several meters out; when the engine began to smoke, it was shut down. ENSCO Inc.'s vehicle, a bath-tub shaped four-wheeler that had flipped during the qualifying rounds after getting caught by a gust of wind, managed to repeat that performance at the first 90-degree bend with no wind at all. [For a complete list of results, plus photos and a plot of the course, see https://www.grandchallenge.org.]

In the end, there were no winners. The vehicle that went furthest, Sandstorm, had been the odds-on favorite. It had the most sophisticated suite of sensors, including a stereo video camera and a spinning 3-D long-range scanning laser seated atop a computer-controlled three-axis gimbal, plus radar, short-range lasers, three GPS antennas, and six Itanium and Xeon workstations; not surprisingly, it had the longest list of corporate sponsors, including Boeing, Intel, and SAIC. Despite a near-fatal (for the vehicle) rollover just days before the qualifying rounds, Sandstorm covered 12 kilometers, or about 5 percent of the course, before getting stuck in a series of hairpin turns at Dagget Ridge. With its left side and wheels leaning out over a steep drop, the vehicle continued to rev its engines until much of the rubber on its front tires had burned off and both half-shafts broke.

Coming in a close second was an Israeli dune buggy fielded by defense contractor Elbit Systems Ltd, based in Haifa. (Grand Challenge rules stipulated that teams had to be led by a U.S. citizen, so Elbit contacted SciAutonics LLC, made up of employees of Rockwell Scientific Co., of Thousand Oaks, Calif.; SciAutonics, which planned to field two robots, agreed to collaborate with the Israelis, according to Elbit program manager Motty Ben-Shalom.) The team, many of them former Israeli Army officers and engineers, had spent a month in the Nevada desert testing its vehicle. But when their robot got stuck on an embankment at kilometer 11, the race for them was over.

Third overall was one of the simplest vehicles: a dark-green Toyota pickup truck customized by David Hall, president and founder of Velodyne Acoustics, Inc., of Morgan Hill, Calif., a maker of high-end speakers. Hall, a robot hobbyist and IEEE member, had spent much of the last year working on his Grand Challenge entry, designing the onboard computer and writing all the code, including the algorithms for GPS tracking and for controlling the vehicle's pair of stereo cameras, which served as its only sensors. Before race day, Hall wasn't optimistic. "Well, it's a good soldier, and if it's told [by its computer] to drive into a ditch, like a good soldier, that's what it'll do. I'm realistically six months away from being able to stay away from that ditch. So if I make it through, I shouldn't have." At things turned out, Hall's robot paused at the 10-kilometer mark to let a tow truck pass, and upon resuming, got hung up on a large rock.

The biggest surprise may have been Golem, a Ford pickup truck that had floundered so badly in the qualifying rounds that it had been slotted to run 14th (with slot 15 going to an autonomous motorcycle). The team's only concern was "to get out of the starting gate and avoid total humiliation," says team member Jeff Elings. The evening before race day, realizing that the robot's sensors (including an infrared camera, two radars, and a gyroscope) were doing the vehicle no good and were probably hampering its efforts, the team stripped them all off. They set the vehicle to run at a steady 40 kilometers per hour, with no braking, and let it go. Using only GPS navigation, the robot drove an astounding 8 kilometers, eventually rolling to a stop on a steep incline.

As for Özgüner's robot, an intimidating 3-meter-tall, 14 500-kilogram acid-green military supply truck known as TerraMax, it proved less monster truck than gentle giant. After getting about 2 km out, it stopped, sensed an obstacle, and then backed itself into some bushes. Sensing the brush as impassable objects, it continued to brake and then roll backward for another few tenths of a kilometer before being shut down.

Though uninitiated spectators may have been disappointed by the day's failure to produce a winner, the event's organizers seemed pleased. DARPA's chief goal in organizing the robotic rally, for which it spent an estimated $13 million, was to advance R and D on autonomous ground vehicles for the "battlefield of the future." At a press briefing following the race, DARPA director Anthony Tether said, "Today was a most important first step in a long journey. Although none of the vehicles completed the course, and we were not able to award the cash prize, we learned a tremendous amount today about autonomous ground vehicle technology. Some vehicles made it seven miles, some made only one mile, but they all made it to the Challenge, and that in itself is a remarkable accomplishment."

The agency has funding from Congress to continue holding the event through 2007, and is rumored to have doubled the prize money to $2 million for next year's race.

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions