DARPA figures out how to run a $2m robot race
Grand Improvement with revised rules
Hoping to make the $2m Grand Challenge live up to its name, DARPA officials have proposed a much more thorough set of qualification procedures and rules for the second running of the race.
DARPA on Monday released its outline for the Grand Challenge II robot race to be held in October of next year. Prospective contestants have been asked to provide their feedback on the rules, which is just one of the welcome changes to DARPA's methods this time around. DARPA has also set up a lengthy but precise qualification schedule to separate the robot gimps from the true contenders, hoping to avoid this year's Grand Challenge I flop.
"DARPA is committed to establishing rules that are easy to understand," said program manager Ron Kurjanowicz. "Feedback from the community is an important part of this process."
Competitors in this year's race thought the Grand Challenge I rules were easy enough to understand until DARPA began changing them late in 2003. DARPA was hit with a flood of last-minute entries in the first race that made it difficult for the agency to decide which teams qualified to compete in the event, which requires unmanned vehicles to race across the Mojave Desert.
In the end, DARPA made a controversial move to limit the field of contestants to around 25 teams that had submitted early papers outlining their vehicles' specifications. Many suggested this gave well-funded, well-publicized teams an advantage over smaller competitors.
For the 2005 race, DARPA has set up a stringent qualification process. Step one of this process requires teams to send in basic team information, a vehicle specification sheet and video demonstration by March. DARPA will then select the teams who qualify for an onsite visit. From there, DARPA will judge competitors' technical papers and then conduct a live qualification event in September.
These methods should up the level of competitors showing up for the race. In event earlier this year, only a handful of teams made it much past the start line with none of these teams making it more than 10 miles.
DARPA has kept the race distance fairly consistent, setting it at 175 miles. The vehicles will still need to pass over natural and manmade obstacles such as ditches, water, boulders, barbed wire and power line towers. The teams will also have to pass through narrow corridors. No vehicle can be larger than 10 feet in width and 9 feet high or weigh more than 20 tons.
Those familiar with the first race will recall that the vehicles are given a series of GPS points just moments before the race begins and must program these into their vehicles as a guide. The robots can use radar, laser radar and other sensing equipment to help guide them around obstacles. The first team to complete the course in under 10 hours wins the $2m prize - double the pay out from this year.
Congress gave DARPA the go ahead to offer a cash prize, hoping to spur innovation in the autonomous vehicle field. By 2015, the US must have one-third of all its vehicles operate unmanned, according a federal mandate. The Grand Challenge event was picked as one way to reach this goal quicker by finding interesting ideas lurking in small companies and universities. Large military contractors have been seen as too inefficient to create the technology needed for autonomous road vehicles.
It's amazing to see DARPA get its act together after largely flubbing the first race and angering many competitors. Many were surprised at how unprepared the agency appeared before and during the race. Hopefully, the competitors will also come to the desert with more gumption. ®
Crash test dummies get date for DARPA Robot Run II 
DARPA doubles cash payout for second robot race 
Robot wars: One man's story of promotional monks and mechanical friendships 
DARPA's Grand Challenge proves to be too grand 
Final robot grunts picked for $1million DARPA race 
Robot grunts tumble in race for $1m prize 
$1 million Grand Challenge map leaked on Web 
DARPA quells robot road rage 
DARPA chisels little guy out of $1 million race 
DARPA's indecision threatens integrity of $1 million race