I, Robot... Soccer
April 8, 2016 12:01 PM Subscribe
RoboCup is a robotics competition featuring autonomous humanoid soccer-playing robots. The action may be a bit clunky at times, but there are some surprising moments. The project's goal is ambitious and maybe a bit threatening for fleshy carbon-based soccer players:
By the middle of the 21st century, a team of fully autonomous humanoid robot soccer players shall win a soccer game, complying with the official rules of FIFA, against the winner of the most recent World Cup.
Woah! This is like the game Goofball Goals in real life!!! They move with the same kind of wonky-ness and fall over for no reason!
Seriously cool, and seriously hilarious. Awesome find!
posted by mayonnaises at 12:09 PM on April 8, 2016 [1 favorite]
Seriously cool, and seriously hilarious. Awesome find!
posted by mayonnaises at 12:09 PM on April 8, 2016 [1 favorite]
Wow, it's weird seeing RoboCup on the blue; I was on the Bryn Mawr team at the 2012 US Open. We placed a respectable fifth! (out of six teams. Well we tied for fifth. So last place. Meh.)
Robot soccer is hard, because bipedal robot movement is hard. The two most important things for any team are a good get-up routine (because the robots fall a LOT; they also damage fairly easily so the company that makes them has repair techs on-site) and a good walk. B-Human, the team that dominated for years and years, did so largely on the strength of their walks.
One interesting thing is that the rules purposely get harder and harder every year to constantly spur innovation. My year was the first year that the goals were the same color - previously goals were two different colors so the robot could tell the opponents goal from its own by color sensing; starting my year you had to do things like directional detection, which is a whole lot harder than it sounds. There were a lot of own goals that year.
posted by Itaxpica at 12:12 PM on April 8, 2016 [9 favorites]
Robot soccer is hard, because bipedal robot movement is hard. The two most important things for any team are a good get-up routine (because the robots fall a LOT; they also damage fairly easily so the company that makes them has repair techs on-site) and a good walk. B-Human, the team that dominated for years and years, did so largely on the strength of their walks.
One interesting thing is that the rules purposely get harder and harder every year to constantly spur innovation. My year was the first year that the goals were the same color - previously goals were two different colors so the robot could tell the opponents goal from its own by color sensing; starting my year you had to do things like directional detection, which is a whole lot harder than it sounds. There were a lot of own goals that year.
posted by Itaxpica at 12:12 PM on April 8, 2016 [9 favorites]
Now my question is, is this the autonomus robotics equivalent of computers winning at chess, or at go?
I believe it will happen within my lifetime, but i'm trying to decide how hard of a problem it is compared to some of the stuff we're trying to get autonomous robots to do.
posted by emptythought at 12:49 PM on April 8, 2016
I believe it will happen within my lifetime, but i'm trying to decide how hard of a problem it is compared to some of the stuff we're trying to get autonomous robots to do.
posted by emptythought at 12:49 PM on April 8, 2016
They move with the same kind of wonky-ness and fall over for no reason!
No reason? Theatrical diving is a standard defensive strategy in soccer. The AI is going for a red card.
posted by dr. boludo at 1:20 PM on April 8, 2016 [3 favorites]
No reason? Theatrical diving is a standard defensive strategy in soccer. The AI is going for a red card.
posted by dr. boludo at 1:20 PM on April 8, 2016 [3 favorites]
Now my question is, is this the autonomus robotics equivalent of computers winning at chess, or at go?
Yeah, it's in the same league. The complexity comes from the fact that robot soccer brings together a whole bunch of areas robots are pretty bad at. Some of it is pure physical motion: stable, fast bipedal motion at anything close to human ability is really tough. Look at all those Boston Dynamics videos: they've had brilliant researchers spending millions of dollars over many years to get bipedal robots anything close to human stability, and they're still nowhere near close on speed. Then there are all the software challenges: soccer requires significant image-processing abilities to find and track the ball, teammates, opponents, and the goals. Then there's programming in teamwork and strategy. All of these are themselves hard problems. And then on top of that, the game moves super quickly and the robots have to keep up with rapidly changing situations, so all that software has to be not only good but extremely fast.
The idea of playing against even crappy, let alone world-class, human players right now seems totally insane. But 2050 is a very very long way away in terms of technical progress.
posted by Itaxpica at 3:04 PM on April 8, 2016
Yeah, it's in the same league. The complexity comes from the fact that robot soccer brings together a whole bunch of areas robots are pretty bad at. Some of it is pure physical motion: stable, fast bipedal motion at anything close to human ability is really tough. Look at all those Boston Dynamics videos: they've had brilliant researchers spending millions of dollars over many years to get bipedal robots anything close to human stability, and they're still nowhere near close on speed. Then there are all the software challenges: soccer requires significant image-processing abilities to find and track the ball, teammates, opponents, and the goals. Then there's programming in teamwork and strategy. All of these are themselves hard problems. And then on top of that, the game moves super quickly and the robots have to keep up with rapidly changing situations, so all that software has to be not only good but extremely fast.
The idea of playing against even crappy, let alone world-class, human players right now seems totally insane. But 2050 is a very very long way away in terms of technical progress.
posted by Itaxpica at 3:04 PM on April 8, 2016
Theatrical diving is a standard defensive strategy in soccer. The AI is going for a red card.
One advantage they have over humans is that when they dive their limbs fall off and loudspeakers erupt with the wails of 1000 infants. They burst into flames which can only be extinguished by magic spray.
posted by justsomebodythatyouusedtoknow at 3:44 PM on April 8, 2016 [3 favorites]
One advantage they have over humans is that when they dive their limbs fall off and loudspeakers erupt with the wails of 1000 infants. They burst into flames which can only be extinguished by magic spray.
posted by justsomebodythatyouusedtoknow at 3:44 PM on April 8, 2016 [3 favorites]
It seems like what we need this for is football—what with all of the head injuries and such. If so, Atari’s CyberBall could have been rather prophetic.
posted by blueberry at 6:33 PM on April 8, 2016
posted by blueberry at 6:33 PM on April 8, 2016
« Older Where have all the gentlemen gone... | "I had an open-door policy on dumb ideas." Newer »
This thread has been archived and is closed to new comments
When someone walked in to remove one of the players and it became clear that they were only about knee-high, it became even more adorable.
posted by sparklemotion at 12:08 PM on April 8, 2016 [1 favorite]