Automatic Chess board

submited by

Style Pass

2021-05-20 13:18:52

In a year of uncertainty and doubt, it was a relief to hear that the 2021 Makerfaire was still on. In 2020 due to the pandemic we were unable to participate in the Makerfaire. As soon as we heard that it was on, we immediately started thinking about what we were gonna make. Our best idea was the Automatic Chess Board. The hardest part about this project was actually figuring out how we were going to implement our ideas. We had an end goal down, but we didn’t have a clear picture of the journey. It took lots of time and rigorous planning and mistakes to get where we needed to be.

For the coding process, our plan was to code a smart artificial intelligence that would be able to look at where the player moved a chess piece and then using electromagnets, motors, arduino, and a raspberry pi, make a move in response. The way that the AI knows where to move is because of its design. At first the AI will only have the rules of how to play chess. Pretty much only how the pieces move and that if you get in checkmate, you lose. At first, the AI will only make random moves, but then it plays and loses. It will progress and learn what moves allow the AI to get farther in the game. This was a very complicated process, but it was extremely important to do this. The way that we are going to teach the AI instead of playing thousands of games against it, is by having it play against itself to quickly learn how to play chess.

In the process of making our project, we used a 2×4 cut with a miter saw. We used a drill press to make the indents so that the motors would fit better. We used a chip for the motor with a 12 volt power source and an arduino. Next, we used an angle grinder to cut the rails to the correct size and shape. After that, we used a vinyl printer to print onto a thin sheet of wood. For the code, we used an excel sheet that had a record of previous losses so that the software could learn from its mistakes and improve over time. For our system to see and read the board, we used Open cb2 through python. We hooked up a raspberry pi to the vision system. One thing that was a big struggle was the vision system. It took a long time and the time crunch we had made it more difficult.