An AI project for solving Jungle board game, using MinMax algorithm with alpha-beta cuts.
To run application the python interpreter in 3.0.10 version needs to be set, with Application.py file (in directory JungleGame\Application.py) set as a script path.
By opening and starting the application it will be possible to view 4 different game modes:
- Player vs Player
- Player vs Computer
- Computer vs Computer
- Research mode In the first mode, you can ask the computer for advice by typing "hint". The research mode allows you to select the values of the two computers that play and let the two computers play N games to then be able to obtain the data that are automatically generated by the application. When we select an AI we can choose its difficulty:
- easy: our AI only thinks about eating the opponent's pieces, it doesn't take into consideration the positioning of the opponent dojo if it doesn't see it
- medium: there is a corrective matrix in the evaluation function that modifies the values of the animals also taking into account their position on the chessboard
- hard: now the corrective matrix is present and changes from animal to animal (e.g. a lion near the water is very strong since it can jump, weaker animals can stay near the allied traps to prevent any raids etc.). Furthermore, the high level is divided into two parts, the first uses a series of matrices created by us, the second one instead uses a series of matrices taken from the following document: Discovering and Searching Loosely Coupled Subproblems in Dou Shou Qi - Joseph Burnett. After selecting the difficulty, you can also select the depth of the search tree.
How to play: Thanks to the UI, to be able to play, just click on the animal you want to move and then click on one of the boxes that are highlighted (which show what are the legal movements that can be carried out by the selected animal), in order to carry out the movement. If during your turn, the letter 'h' is pressed, it is possible to ask our AI for help in order to make the best move. To be able to cancel the selection of the pawn, simply click on it again.