DARPA's indecision threatens integrity of $1 million race
All is not well with the $1 million Grand Challenge event sponsored by DARPA, as some participants in the contest have dished a healthy amount of outrage at organizers over several last minute rule changes.
On Nov 6, DARPA (Defense Advanced Research Projects Agency) officials will announce that they have been forced to alter the Grand Challenge rules due to an overwhelming number of entries. The contest, which will have robotic vehicles race across the Mojave Desert, was originally designed to let small companies, engineering enthusiasts and universities show that they could out innovate large military contractors. Now, however, some of the smaller teams are charging that DARPA is set to tweak the contest in favor of well-funded teams - a move that would be a perversion of the race's original tenets.
"They are not being true to their word or the spirit of the contest," said Warren Williams, a member of Team Phantasm. "This was supposed to be about the mom and pop teams having a chance to compete and coming up with new technology. Now, it's the same people as the military complex."
The Grand Challenge event is the military's attempt to spur robotic vehicle technology. Large contractors have for years been trying to create vehicles that can see and steer on their own and traverse the most rugged of terrain. But these efforts have met with little success.
This is a problem for the Department of Defense (DoD), as a Congressional mandate calls for one-third of all military vehicles to operate unmanned by 2015.
To speed things up, DARPA decided to create a contest open to the public. The organization will hold a race in March of 2004 in which the teams will travel 200 miles from Barstow, California to Las Vegas. The teams will be handed GPS coordinates two hours before the race, program these into their cars and then turn the vehicles over to DARPA. Whichever team completes the course first in under ten hours wins the seven figure bounty.
Preparations for the contest had been going just fine until last week, when DARPA officials began rattling contestants' nerves.
The teams were to submit a technical paper by mid-October that would serve as the key to their entry in the contest. If DARPA judged that the paper and technical plans were feasible, then the team would be offered a spot in the March competition.
As the Oct. deadline drew near, however, the number of teams applying for the event jumped from around 50 to 74 and then finally over 100. DARPA was overwhelmed by the number of entries and issued the following notice to competitors on Oct. 28.
"While the response has been more than anyone could have hoped for, there are factors beyond DARPA's control that limit the number of vehicles that can participate in the Los Angeles to Las Vegas event," a DARPA official wrote. "The terrain in which the Grand Challenge will take place is long and difficult. The need to comply with environmental regulations, ensure the safety of the participants and spectators, and complete the event within the number of available daylight hours limit what can be accomplished in one day. Given these constraints, our analysis leads us to believe that only 20 vehicles can be allowed to run the Grand Challenge route on March 13, 2004."
Cutting the number of teams to 20 was a blow but not unexpected. The more challenging news was to follow.
In the same Oct. 28 letter, DARPA informed the teams that it would visit each site to determine who could compete in March. "To be fair," it would select a total of 25 teams and have them perform tests the day before the race, letting the final 20 race for $1 million.
One week later, however, DARPA dealt another blow to the competitors.
DARPA said it has already selected the 20 teams that will be able to race in March. The remaining teams whose technical papers were judged "Possibly Acceptable" must request an onsite visit from a DARPA official and then fight for one of the six remaining spots.
Smaller teams see this decision as a problem for a couple of reasons.
Many of the teams had been waiting for their technical paper to pass muster in order to receive funding for their projects. Sponsors were hesitant to hand out cash or equipment before they knew the horse they were backing would be able to have a shot at the race. Well-funded teams such as those at Carnegie Mellon and CalTech already have incredible resources at their disposal, including help from large military contractors.
The teams need money for their vehicles - designs that range from Hummers to ATVs - and a host of other technology, including sensors, radar and laser radar. In many cases, the contestants also need access to satellite imagery, GPS testing gear, computers and whatever specialized technology they have come up with.
"The same guys DARPA has given money to in the past are now the ones competing in the race," Williams said.
Others share Williams' opinion.
"The real work in robotics is done the way the first airplane was brought into being: by T&E engineering in basements and garages; primarily by unknowns with whatever can be scavenged from wrecking yards, scrap yards, surplus sales, and dumps, augmented by a handout from engineers at corporations here and there in the form of samples of (electronic) components one could not otherwise afford," wrote one competitor in a letter to all the groups.
"It is objectionable that DARPA make such drastic changes in the qualification process less than 5 months before race date," wrote a member of the Arctic Tortoise Team from Alaska. "This is especially true in light of the fact that attendance at the Feb. Competitors conference was FAR greater than anticipated. This should have provided ample warning for DARPA to address these issues in a timelier manner. . .It is clear that the recent rules changes will disproportionately affect the smaller, more innovative teams, and thereby favor more conventional approaches. This in turn, compromises the stated mission of the Grand Challenge event."
Team Arctic Tortoise, one of the 20 already approved to compete, has considered withdrawing from the event, in part, to "protest the flippant disregard for the competitors’ significant investment."
A lively debate on the subject can be found here.
The well-prepared teams claim that if a vehicle is not up, running and being tested now, it doesn't really stand a chance of winning the event anyway. Anyone affected by DARPA's last minute change simply wasn't on the ball.
For many teams, however, the news is devastating. The contestants thought the technical paper was just the first major hurdle in an endurance test. They would secure their spot in the race, use that to drum up support and then work for months on their vehicle.
This is a painful chapter in an otherwise glorious affair. DARPA has found a way to save taxpayers millions of dollars by using human nature and the spirit of competition - instead of a lucrative military contract - as a motivator. The teams are spending their own time and money on their creations.
The vehicles these teams design may be used to rescue soldiers, perform reconnaissance and even fight battles. This sounds far more useful than the Terror Casino and Total Information Awareness projects proposed earlier this year by DARPA.
Now, however, DARPA will be relying, to a large extent, on the same minds that have failed to create robotic vehicles in the past. Most of the participants interviewed do not feel that any team will win the race this year. That should be the best signal to date that DARPA needs to keep its focus on the unknown and let as many teams as possible have a crack at using their imagination. ®
Sponsored: Optimizing the hybrid cloud