scrapbook

TECHBITS: pool-playing robot

KINGSTON, Ontario (AP) -- Deep Green, a pool-playing robot built in a lab at Queen's University, isn't spending nights at bars challenging human champs -- yet.

Michael Greenspan, an associate professor of electrical computer engineering, sees the project as a way to explore robotic vision and game theory. But it began with less lofty notions, when, after a frustrating night of pool at a robotics conference, Greenspan decided it would be easier to build a pool-playing robot than master the game himself.

"At the point the system beats me, I'll have been proven correct," Greenspan says.

So far, Deep Green breaks about as well as its creator, but needs to work on accuracy. To that end, Greenspan and his students are developing better machine vision algorithms. They're also working on the computer's strategic abilities, an area where Greenspan sees the possibility of spin-off technology, since pool is more uncertain than most games.

"In chess, there's no uncertainty. A piece is either on a square or it isn't," Greenspan says. "Pool works in a continuous world."

[original article]

Posted by thinkum at August 18, 2004 04:45 PM