About the Project

Created an autonomous reversi playing agent enabled with classical AI search algorithms. The search algorithms used are minimax search with Alpha-Beta pruning. The agent is triggered after each turn played by the opponent player. The bot makes the next move based on the rules of the game, as well as the next best scoring turn for the bot. The bot constructs the minimax tree, and does an n-ply search using Alpha-Beta pruning.

The heuristic score is adjusted according to best strategies in the game, especially having high values for corner tiles. The heuristic function is constructed keeping the stability of the algorithm in mind. The best move is made by the bot, after accumulating the list of next moves possible and choosing the one with the best overall outcome, based on the heuristic function. This bot won the 2nd prize in tournament mode among almost 20 bots, beating 18 other bots, mostly with high win margins.

Also explored other algorithms like A*, SSS*, before settling with AB pruning, based on performance of all algorithms.