Slashdot Mirror
Kiro, the Foosball Robot
JasonFleischer writes "Although the official line from the RoboCup competition is that robots should be ready to challenge humans on the soccer
field by the year 2050, we don't have to wait that long to see man-machine
competition in the bar.
Researchers in Germany have developed a table football (foosball, table soccer, whatever) robot. The human challenger(s) take the red team, while the machine works the blue side, using an overhead video camera to see what's happening on the table. The conference paper shows that while the machine generally wins against the normal bar-amatuer it has no chance against a human grandmaster. But these kinds of things are always improving, after all look how big a deal the man-machine chess competition turned out to be. So perhaps the current table football world champions should be watching their backs."
This will be fun for about ten minutes, until the machine is either impossible, or laughably easy (more likely) to beat.
People in bars want games they can play socially, with other people. The people who would really be in to something like this would stay home tweaking their Debian installation and picking at their hemorrhoids.
One of the main strategies is to perfect a shot that can be executed faster than humans can react to it. (i.e. stop the ball with your offensive 3 man rod and then wait and then slide the ball and shot it somewhere else along that rod). But what happens when the computer it watching it, if video is fast enough and the ai sees what you are doing then it might be able to block things that humans would never be able to stop. On the flip side the robotic arms could execute the same shot 100% of the time, but I don't think that the overhead camera would give it enough information. There would also need to be sensors in the motors that move the rods because that feedback is important.
(Maybe we could put a GPS device in the ball!)
As much as it might be true that it'll have a better win/lose ratio with sensors, GPS is not the way to go. I mean... A foosball table isn't all that big, and what will the computer do when he thinks the ball is actually 5 metres outside of the "arena"? GPS is a no-go.
I'll be happy when they teach AIBO to fetch the morning paper or autogrowl at bad people. I wonder if they'll get to a point were I can download new "programs" for the robot -- like in The Matrix -- download a martial arts foosball playing AIBO program. Now that would rock.
SPAM solution made easy: 1 spammer, 5 cords of rope, 5 hourses, and fireworks. Be creative.
I think everyone so far has missed as to why exactly this is such a big deal.
It is not important what game the robot mastered. What is important is the fact that the robot is capable of on the fly visual/apendage (arm) coordination.
That means we have technology that will allow for application of this in other areas (think retrieval robots for disaster zones, etc.)
That is what makes this story so exciting.
It is a research project. They're doing it to see if they can do it. Creating a robot that cn play foosball is obviously an interesting technical challenge in a lot of ways. The researchers will learn a lot in the process.
Words to live by, my friends.
You were 80% angel, 10% demon. The rest was hard to explain. - Over The Rhine
"Math in a song is good."-Linford
Having played foosball my entire life, I doubt they can make something to beat anyone of real skill anytime soon. When you get really good at the game, it's not about seeing the shot to block it, because you rarely see it (if you're good enough) because it's so fast. No, you're going on anticipation, of what the player may do. It becomes a sort of guessing game at that point.
This is where the stragegy in foosball really comes in. They key is to have a bunch of various shots you can use but that all "look" the same on set up, so the player has no idea where you're going to go with the shot. If they guess wrong, you score. Of course, this takes a lot of skill to do well (and just as much skill on defense to defend).
Because the robot is using a camera, and because the action is so fast, I can't imagine it stopping a good push shot, or even a good pop or bank shot for that matter. Similarly, I can't see it setting up good shots that a human couldn't pick out everytime. Though I think it would be easier to teach a robot offense, than defense.
No really, I'm serious. This robot's playing style is more akin to what we call "whackball." Better foosball players remain in contact with the ball a lot longer, "palming" the ball with the face of the man to do quick changes in direction, fakes, and more. There's some video available here on the right side of the page.
Of course, there have been robots interacting with dynamic environments in similar ways for a long time, such as juggling and running. It's a big jump to go to the next level, which requires chaining together sequences of difficult actions, such as palming the ball, passing, and shooting. But I think foosball is a great place to explore such dynamical interaction and action composition, and I'm jealous that they beat me to building a table.