The version for the contest had an extremely poor AI. The logic went like this. If the computer could capture a piece, it would do so. If there were multiple captures, it would randomly select a piece to capture. If there were no pieces to capture, the computer randomly selected a valid move. Pretty lame. Especially considering it would often land right into a capturable location. But that was because I did not have enough time to create a decent AI. I turned in what I could get done. I was just happy the computer actually MOVED!!! That was my main goal for the contest, otherwise, there really would have been no game.
I now have a much better AI. It has beaten me on several occasions. Although some of those are just because I am purposely testing the computer AI. In any case, the AI is much improved and I hope to release an update sometime in the near future.
In my quest to understand computer AI, I have been trying to understand the Minimax algorithm. I already base the move on some type of ranking system, but the ranking calculation is not fully implemented. It is currently based mostly on what pieces can be attacked and whether or not the computer can be counter-attacked. It's actually fun and does require some thinking to beat.
Next steps:
- Work on creating a better way to rank moves.
- Implement the Minimax algorithm.
- Create a 3rd map
- Update the avatars for pieces.
No comments:
Post a Comment