Am I the only one who doesn’t think that the fact that the DARPA Grand Challenge being completed this year by several teams, compared to last year’s complete failure by all teams, is that big a deal? Professor Reynolds thinks it’s evidence of “sudden and dramatic progress,” and The Speculist weighs in with similar thoughts. I have to disagree with those analyses. The original Grand Challenge teams from 2004 had, I believe, less than a year to gather funding, assemble a team, and prepare a vehicle and GNC system, which is a lot of systems integration work. That would be an impressive effort by anyone, and the fact that no team was able to complete the 2004 Grand Challenge comes as little surprise to me.
As far as showing that there is rapid advancement, the International Aerial Robotics Competition has shown that the appropriate technologies for autonomous navigation exist and have existed for several years, with the Georgia Tech team showing Level II behavior in 2003, meaning the aircraft was able to fly into a simulated village (at an Army MOUT training center), find a particular building, and find an opening in that building. Further, the Mars Exploration Rovers have image processing technology that can perform off-road driving, and the IARC guys have been able to do basic navigation for several years, with the current course being 3 km one way—a 6 km round trip. So again, the technology exists, this is just one more (albeit highly robust) application.
Other things worth noting: the Volkswaken Touareg, such as the Stanford entry used, has 71 ft3 of cargo space with the rear seats folded down. In tandem with an upgraded electrical system, this is more than enough for some really bitchin’ computer power. Real-time image processing is not a problem for them. But it wasn’t a problem last year, either, and with that much space could have been done in 2001 or 2002, again, given about the same preparation time.
While the DARPA contest is a good thing, and while I’m sure it is difficult, it doesn’t seem to me like anything terribly revolutionary. I expect that there will be a change in the Grand Challenge that makes it more difficult, much like the IARC did when their previous challenges have been completed, and that it will prove to break systems that worked fine in this years challenge. It had better, or it’s clear DARPA has set the bar too low.
What the Grand Challenge 2005 does is provide a baseline. Now the top-tier participants know the rules, and they have a working, reasonably reliable system, that I hold could easily have been built several years ago, given they were provided about the same amount of time to build it. How these systems are leveraged and improved in Grand Challenge 2006 and beyond will be the true test of how fast our capabilities are progressing.
