DARPA 3Future Nobel Prize winner Brian Pilnick did see the post about DARPA 3, and here are his comments:
My first thought when I saw the announcement yesterday was that it is going to fail miserably. As cool as our tech is, its just not at the level of safely merging with moving traffic, negotiating busy intersections and tight urban environments.
That being said however, there are a few reasons why it could work.
First off, teams have a full year and a half. Second, DARPA will be funding some teams with a million dollars. It wasn't very publicized (not much of the Grand Challenge was) but there were a couple teams that had they finished, would have destroyed both CMU's and Stanford's times last year. One was ENSCO, who's a relatively big company, but the other, if I remember correctly, was a small group that had nothing to do with robotics. If a team like that had an extra million to work with, they would be a tough contender, and that's what the original intent of the Grand Challenge was when it was started several years ago--to allow garage engineers to compete on this level.
It could also work because if you do the math (60 miles in 6 hours), you only need to average 10 miles an hour, which is really slow, so the tight area and obstacles become much less of a problem. The final reason I think it could work is DARPA themself. This past challenge, they showed they were willing to greatly sacrifice the difficulty of the problem in order to declare a winner. Even more than actually getting the technology, they're interested in getting people excited about the technology and getting money flowing into the problem.
Either way, it will be very interesting to watch.
Brian also let me know that the NOVA documentary on the 2005 Grand Challenge is now online at 2005 Grand Challenge.