DARPA Urban Challenge

I really ought to finish this entry. Better late than never, eh?

Yesterday (being about a week and a half ago), I woke up at 3:30 AM, which was strange, because it was a full half-hour before my alarm was set to go off. I needed to get an early start on the day if I was going to be in Victorville (over 100 miles away) by 7:00. On a related note, dawn is a lot prettier when it comes in the morning (every other time I remember watching the dawn, it has been while taking a break from work which I’d been doing for the past 12 hours).

I went to the final match of the DARPA Urban Challenge. You may remember that in 2004 and 2005, DARPA held what was called the Grand Challenge, a contest to see who could get an autonomous vehicle to navigate ~150 miles through the desert without getting stuck somewhere. No one finished the race in 2004, but in 2005 Stanford narrowly beat Carnegie Mellon (and some other people finished too).

The Urban Challenge was a similar idea: get the cars to autonomously drive many miles. However, this time it took place in an urban environment: the cars needed to stay in their lanes, stop at the stop signs, and not hit the dozens of other cars driving around the area at the same time. The winner would be chosen based not only on speed, but on good driving techniques: ideally, the winner would be able to pass a drivers license test.

These extra cars were pretty neat: they had everything except the driver seat ripped out and replaced with steel bars and cages. They were driven by professional drivers (on a closed course, no less!). These precautions appeared to be necessary: at least one car had a large dent in it by the end of the day.

The event started off horribly bureaucratic: some DARPA official gave a speech about how patriotic it was to develop these vehicles, so they could help the military and aid our troops. This was followed by a small parade of marines on “wild mustangs,” which were the tamest horses I’ve seen in a while. The podium where they presented the flag and sang the National Anthem had a bunch of statues of bald eagles on it. However, after this junk, we got some robotic cars to drive themselves around, which was pretty cool.

There were ~30 semifinalists as of a couple weeks ago, who were narrowed down to 11 finalists last week. Of these, about half were disqualified or otherwise removed in the first half hour of the race. However, I believe all the remaining cars after that finished.

There were at least 2 crashes: 1 I know about because one of the professional drivers had a big dent in his car. I also got to see the latter half of a very interesting crash: there was a Y-shaped intersection, and the idea was to drive up to it, stop at the stop sign, and take the right fork. The Cornell car drove up, stopped at the sign, turned too far, noticed the curb coming up, stopped, paused, and backed up to try again. After some more thought, it turned again, turned too far, saw the curb coming up, stopped, backed up, and paused for a minute before trying again. In the meantime, the SUV from MIT had come up behind it, and was getting impatient. It crept forwards, and eventually decided to switch lanes and pass the Cornell vehicle, turning smoothly into the correct lane. However, just then Cornell got it right, and the two cars hit each other sideways. It was quite exciting.

Stanford finished fastest, followed closely by Carnegie Mellon and Virginia Tech. However, by the next day, the judges had decided that the CMU entry was a better driver, and so it won. Stanford got second place, and Virginia Tech got third.

Other interesting things: I got to shake Sebastian Thrun’s hand, gave him my business card, and mentioned that I’m very interested in the Stanford robotics program. For those of you who don’t know, Sebastian Thrun has been one of the foremost roboticists in the world for over a decade. He also invented Monte Carlo Localization. It was pretty cool.

Leave a Reply

One Comment

  1. hmcmodelt says:

    My adviser is on the MIT team and he gave us a review of the whole race from MIT’s perspective. That Cornell-MIT incident was officially ruled as a no-fault accident, leaning towards Cornell hitting MIT. And there was another incident where MIT and another car hit because the other car was driving the wrong way down a one-way street.

    That’s so cool that you got to go see the race, sounds like it was pretty exciting.

Leave a Reply to hmcmodelt

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>