Hi all,
I’m working on a chess-playing robot called Gambito. The name means “little gambit player†and alludes to the Intel/U. of Washington research robot called GAMBIT. Gambito aims to mimic and rival GAMBIT’s ability to play chess autonomously, using low-cost or open-source components.
Construction
The arm is made using the AL5D hardware, with a longer forearm for a longer reach which is needed to place pieces at the far end of a 15x15 inch chessboard. The software is a custom Python program which I plan to release to open-source soon.
Below is a demo video. Let me know what you think. For more info, visit:
www.borglabs.com/projects/gambito
Enjoy,
Ben Bongalon
Wow that’s one cool chess playing bot! 
What was the length of the new forearm tube? I just ordered 8", 10", and 12" tubes to add to the SES. You might be able to increase the repeatability a lot by creating some mechanical alignments for the chess pieces. This could be done with small magnets added to the underside of the chess board, with metal parts embedded into the bottom of the pieces. Just a thought. Looks great…
Thanks for the kind words, Jim. I used an 8" tube, its a perfect size because anything longer would require extending the shoulder arm as well and probably need beefier servos. I think I found the main cause of the bad grasping which happens mostly when the arm reaches out to the 2 rows farthest from the board. The culprit appears to be the analog servo that I used for the base rotate. I used an HS-485HB which has a 8us deadband. This corresponds to at lleast 0.8degrees of slack (ie, 180degress * 8us/(2400-600)) and from my experiments, it seems more. When the arm is fully extended as it would be when reaching far, this slack translates to sin(0.8degrees)*17inch = 0.24inch of play which too inaccurate. I will buy a digital servo to see if that improves the reliability. But I don’t think this will fully fix the problem, just reduce it.
Yeah, the real solution is too ensure that the gripper is aligned to the target. The magnet idea is pretty cool however I’m obsessed with using a webcam (placed between the two pincers of the gripper) to hone in on the target. I’ll give it a shot… if it becomes too hairy I may just do the magnet approach.
Sounds good. Looking forward to seeing how it progresses.
What was the length of the new forearm tube? I just ordered 8", 10", and 12" tubes to add to the SES. You might be able to increase the repeatability a lot by creating some mechanical alignments for the chess pieces. This could be done with small magnets added to the underside of the chess board, with metal parts embedded into the bottom of the pieces. Just a thought. Looks great…