Slashdot Mirror

← Back to Stories (view on slashdot.org)

DARPA Robot Contest Update

Posted by michael on from the hunter-seeker dept.

rbrandis writes "DARPA has selected a wide variety of teams, after a series of last minute rule changes and a solid outpouring of anger, the final list of competitors for DARPA's Grand Challenge robot race has been set with 25 teams preparing to try and win a $1 million prize." The anger is exemplified by submissions like this one: Totally_Lost writes "Last spring we flocked to DARPA's Grand Challenge media event in Los Angeles to be told that they wanted everyone's participation in their Robot race this March. They told us that the race would be open to Mom and Pop garage sized participants - and Lied. This fall, nearly 100 teams completed technical paper submissions, with about half to be eliminated from the $1M prize race because they were too small to be 'real' competitors. Well, the rejected robot racing teams got together in Las Vegas last month, and formed the International Robot Racing Federation. This month IRRF is announcing its first competition with $1M in prizes pledged by sponsors, and lesser prizes too, to be offered in a REAL OPEN Challenge next September (providing the race that DARPA failed to deliver)."

6 of 142 comments (clear)

  1. DARPA not open in its policies, i'm shocked. by jester69 · · Score: 5, Funny

    They have always been heretofore so up front and honest. This is truly a departure for the military industrial complex.

  2. It looks like DARPA wins... by mellon · · Score: 5, Insightful

    ...beyond their wildest dreams. Not only do they get to have their own competition, which may produce some interesting results, but in addition they get to see another competition that they don't have to pay for, and if anything cool comes of it they can always step in and make an offer on the technology. Plus, a new hobby is born. Sounds like everybody wins here.

  3. Re:Join Team Overbot - no pay, some risk, big priz by Anonymous Coward · · Score: 5, Interesting

    I'm pretty good at smelling a troll, and this one reeks. A quick search of Google shows us that this exact same comment was posted here over 2 months ago. The parent is probably using the Anti-Slash Database Tool to find and search out previous popular posts.

    In addition, the parent is a brand-new account, with an already rich posting history of highly-moderated comments, some of which are reposted from older articles.

    Also, if you carefully read between the lines, you will notice the posts by this user bear a striking resembelance to those of Sir Haxalot, Pingular, and Steve 'Rim' Jobs, all of which are accounts created by the same user for the purposes of karma-whoring and building up large amounts of karma very quickly in an effor to use this to his advantage while trolling.

    Please moderate his post down so this trolling karma-whore will not be able to annoy others and carefully work the system. If you fear the wrath of Meta-mod, you can always rate him as "Overrated" which gives negative karma but does not go to M2.

    This has been a public service announcement from a helpful Slashdot user. Posted anonymously to avoid the groupthink.

  4. Just like WWF by AndroidCat · · Score: 5, Funny

    All sorts of heated tempers and a split to a rival federation. All they need now is a few good rants, some cage matches, and one bot hitting another with a chair or something. It'll be a shoe-in for weekend afternoon TV. w00t!

    --
    One line blog. I hear that they're called Twitters now.
  5. If you don't have your own machine shop ... by inonit · · Score: 5, Interesting

    You may want to check out RARS, a simulator framework in which you can write programs to run in a simulated auto race against other programs. I haven't messed with RARS in a while, but at the time I was using it, (IIRC) your driver was a C++ class that received a huge struct as a parameter and returned a small struct indicating the direction you wanted to steer and a number indicating gas/brake magnitude.

    But what do I know -- my car could barely make it around the track without running into the wall.

  6. The real issues Grand Challenge competitors face by Animats · · Score: 5, Informative
    The real things competitors are complaining about are these:
    • This is turning into a breadcrumb-following exercise. DARPA provides a "waypoint file" on CD-ROM two hours before the race. For each segment between waypoints, there's an allowed path width, indicating the area in which the vehicle must stay or be disqualified. Originally, the vehicles were supposed to be truly autonomous, with the DARPA-supplied waypoint data providing only general corridors. That's what made this interesting. Then DARPA said they would provide about 1000 GPS waypoints. Now they're saying it will be about 5000 GPS waypoints. With 25 waypoints per mile, it may be possible to do this on GPS alone, blindly driving from waypoint to waypoint, with some minimal obstacle detection and avoidance. That's not "autonomous". That's preprogrammed, like the old Milton Bradley Big Trak toy.
    • Some teams are using a "semi-autonomous approach". In the two hours between the release of the waypoint file and the start of the race, large numbers of people at remote sites will manually plan out each segment of the trip, using aerial photographs and maps. The trip segments will be combined, downloaded to the vehicle, and used to drive it. DARPA has approved this approach. That's not "autonomous", either.
    • Government funding of entries has been a big issue. Caltech is using the "Perceptor" software package, which fuses overhead and ground imagery. Perceptor was developed with DARPA funds at JPL, and is not available outside JPL/Caltech. The Caltech team formally asked DARPA if this was OK, and DARPA said yes. Other teams complained. The head of the CMU team is currently a principal investigator on NASA's Hyperion robot program, which raises some red flags about Government funding. DARPA is now requiring teams to provide a "certification of self-sufficiency" statement, with lines like "If the hardware and software is proprietary to my team, it was not developed or purchased using U.S. Government funding either directly ... or indirectly". CMU has received Government funding for robotics work for decades, and it's not at all clear if any of that crept into their entry.
    • We're hearing rumors that the 2004 event might be the last one, even if nobody wins. This is apparently an internal issue within DARPA, and we haven't heard details. DARPA's officially stated position is that the event will be held "approximately annually" until someone wins through at least 2007, when the Congressional funding runs out.

    The selection process wasn't hard for anyone who had a clue. DARPA was evaluating papers for months, and you could resubmit as many times as you wanted. DARPA warned entrants in the rules that it might take several turnarounds to get a paper through. The people whining about rejection submitted papers at the last minute.

    We'll be in Fontana in March.

    John Nagle
    Team Overbot