Tuesday, 23 November 2010

Chess playing robots - an update

About 3 years ago I posted an article on this blog about the potential development of a real chess playing robot. In the article I suggested it might become a reality in 3 to 5 years time. This week there was the news of Vladimir Kramnik playing blitz against a robot, so it looks like it is getting closer.
However I say 'closer', as opposed to 'here', as the setup for the Kramnik game fell short of what I had envisaged. Based on the news report and pictures, it looks like the game was played on a DGT board against a robot arm. I'm guessing the board transmitted the moves to a computer running a chess program, and the computer then told the robotic arm controller the actions to take in reply. Visually impressive, but on one level, fairly simple (not saying I could have built such a system btw).
What I would like to see is (a) a system that can play on any kind of board using vision processing to determine the moves (b) a robotic arm to play the moves, press the clock, and set up the pieces for the next game and (c) be mobile so it isn't restricted to one specific location in a tournament. And for extra credit, a voice recognition and response system that allows the program to analyse at the end of the game.
Not much to ask for really, although my meagre attempts at part (a) haven't had good results so far.

No comments: