Artificial Intelligence

Dao is a two player abstract strategy board game. It comprises a 4*4 playing board with four playing pieces for each player. The aim of the game is to arrange the pieces in certain winning formations before the opponent can. It is unique because any piece that is moving must move as far as it can. There are four different ways to win the game. One is by forming a 2*2 square with four pieces, by forming a straight vertical or horizontal line with your four pieces, placing your four pieces in all four corners, and by having one of your pieces cornered by three of the opponents pieces . The aim of this project is to create an artificial intelligent agent who plays this game with the human or with another agent. Visual studio environment is used to write the agent. In this project I tried different algorithms but working algorithms is greedy search.

First Tourney

For the first tourney, the agent generated all possible valid moves for the current board state. Vector data structure is used to store all possible valid moves and all possible valid board states. Random function is used to generate random index number and this number is used to pick the random move from the vector which has all valid moves. In the first tournament the agent played well compared with other agents, but it wasn’t using any of the intelligent moves. It just picked the random move from available moves.

Second Tourney

For the second tourney, first I was planning on using alpha-beta search to detect the best possible move for the current board state. Do to that, I created a tree node including integer variable to store winning weight and vector data structure with tree node data type to store the board states. Here also I used vector data structure to store multiple children. Current board state is stored in the root of the tree data structure and possible valid moves are stored as children of that root. I could