Robocup 2002 Now Underway 182
ChenLing writes "Forget small robots on wheels playing soccer...193 teams from 30 countries have built androids to play soccer at 'RoboCup'. Although as the article says, it will be a while before we can pit androids vs humans on the same field....." Our previous story has background links and information about the contest.
Robocup special on PBS.. (Score:4, Informative)
some of the things the teams could do were amazing...while most teams basically used "brute force" to push the ball into the net, there were some advanced teams that could actually recognize teammates and pass the ball to them...that was pretty impressive...the had kind of a pinball type flipper kinda thing that would poke out near their feet...
another cool thing was that some of the robots (i think it was the same team that could do the advanced passing) could actually "trap" the ball...they had a spinning bar on their front, so as the ball came to them, the bar would spin, and they could keep control of it, by basically continually spinnning the ball towards themselves...kinda like a tread mill...
another progressive idea (and i'm sure the slashdot/open source community will appreciate this) was that after the competition is over, all of the teams are required to share their source code, so everyone starts on the same plane for the next year of research... it's really a cool speical if you can catch it while you're flipping the channels sometime....
I think you are overestimating what is being done (Score:5, Informative)
According to the Official 2002 Robocup Humanoid League Draft Rule [robocup.org] there are three catagories of Current and Future events with several sub-sections. Here is the run down.
* Standing Still on One Leg
* Humanoid walk - out from one end, around a pylon and back,
* Shoot - where the bot is able to shoot on the goal and get it in.
* Penalty Shootout
* 1v1, 2v2, 3v3 Soccer
* Freestyle - Five minutes of judged performance art.
According to the organizers they are just hoping to get some teams to try the first few! And as you can see the competitive playing of soccer is waaaaaaaaaaaaaaayy down the list and probably aren't even being attempted this 'Cup. Sure, those nice posed shots of those Sony bots look nice, but those aren't competition pictures. Sure, the information on the Official Robocup site [robocup2002.org] is sparse, but don't you think that they would have some big announcement if it did?
This article from MSNBC is confusing one league (the humanoids) with another (the non-humanoid) in an attempt to create hype. Personally I think Robocup deserves it, but not by misconstruing what's going on.
The best thing about the Robocup site is that you can actually watch the replays of the simulation games with Flash [uni-koblenz.de]. Pretty sweet.
That's our team! (Score:3, Informative)
Our current robots can trap the ball and move from one end of the field to the other, backwards in under a second. The ball looks like it is stuck to the robot! On top of just pulling the ball in, they also have a transverse dribbler so that the robot can also move side to side without loosing the ball at the end of the roller.
Most teams are pure CS majors, but Cornell uses a team of Mech. E., EE, and CS students. By doing this, the team can concentrate on developing more innovative robots with new features to better play the game.
Oh, and brute-force is forbidden by the rules. A ref. is supposed to assess penalties against robots that hit into the opposing team's players.
-- Len
Re:Wow. (Score:3, Informative)
Usualy the code is divided six ways : vision, mapping, odometry, decision making, real-time sensing and real-time motion commands.
The first three set us the real robots from the simulation. No mater are much simulation leaged proud themselves about dealing with noise - gaussian noise is nothing like what Mother Nature throws at you through real sensors : shadows, glare, obstruction, tripping, pixellation, ccd noise even the crowd. All those can throw your system off on a wim.
The decision making is as easy as your base three module is solid. Its architecture gets a bit tricky though. At McGill we eventually came up with a small bytecoded language to ease the job, and ran an interpreter an the dogs. Cool stuff.
The first year competing with the Aibo, walking was the bottleneck. Whoever could walk to the ball won. The year after, the walking had gotten better, and localisation became key. Every could walk to the ball, but only the better robots had tracked their position well enough to know which direction to kick it to when arriving. In 2001, speed and reliability of the whole made the cut.
And to answer your question : yes, the Aibo stubble over all the time. It makes a good shows when compared to the larger robots which are extra careful not to run into each other. Nobody wants to break anybodyelse's research baby.
Moreover, since the Aibos are equiped with gyroscopes, getting yourself up is only a mater of : if (gyro.vertical() > 45degs) { bring the legs in, bring the legs out }
The dogies have miniature ccd camera and couldn't really tell team appart. The large wheels robots, which have full sized frame grabber, can. Acting on the information is harder. Even if you see your teamate, you have to be able to mesure speed in order to complete a pass - and that adds a source of real-world noise to your system. Few teams manage to get it right, but it gets better every year.
They are certainly more than automated kickers. Better, it does wonderful thing for robot research and is alot of fun to participate in.