James Stanley blogs on the design of automatic chess boards:
Lichess has recently released a Boards API, allowing anybody to interface a physical chess board with lichess in order to play online games using an automatic board. I previously thought about implementing my Strange Chess Clock using an automatic board, but never got round to it. With lichess now supporting online play with an automatic board, it’s hard to justify not making one!
James writes that there are only 2 things an automatic chess board really needs to do. It needs to be able to move the pieces on the board, and it needs to be able to detect where the player has moved the pieces on the board. You can view it as an input/output device: take input in the form of the player’s moves, and provide output in the form of the opponent’s moves.
Complications can arise in certain designs, because in addition to basic piece moves, you need to be able to input and output:
- knight moves: knights are able to jump over other pieces, you need to account for this
- captures: a piece gets removed from the board and the player’s piece takes its place (in particular, if you only measure square occupancy, you might not know which piece was captured)
- castling: the rook needs to jump over/around the king
- pawn promotion: as input, you need to detect which type of piece the player promotes to; as output, you need to grab a captured piece and place it on the promotion square
See the approaches that have been done and how one can look to automate designs in the post here.