zedrichu / chinese-checkers-ai

Implementation of the Chinese Checkers game and an AI agent for it

Chinese Checkers Game AI

This project implements a Chinese Checkers game that can be played in various modes, including against different types of AI players and a human player via a graphical user interface (GUI). The game supports command-line arguments to specify the player types and, for AI players using the Minimax algorithm, the depth of the search.

Installation

Before running the game, ensure you have Python installed on your system. This game has been tested with Python 3.12+.

Clone the repository to your local machine:

git clone https://github.com/Zedrichu/Chinese-Checkers-AI.git
cd chinese-checkers

Usage

To start the game, navigate to the project directory and run main.py with Python, specifying the options for the first and second players.

python main.py --first-player <player_type> --second-player <player_type> [options]

Player types

• human: A human player using the graphical user interface.
• random: An AI player that chooses moves randomly.
• nonrepeatrandom: An AI player that chooses moves randomly without repeating the last move.
• minimax: An AI player that uses the Minimax algorithm with optional depth specification.

Options

• --first-minimax-depth <depth>: Specifies the depth of the Minimax search for the first player. Only required if the first player is minimax. Default is 6.
• --second-minimax-depth <depth>: Specifies the depth of the Minimax search for the second player. Only required if the second player is minimax. Default is 6.

Examples

Start a game with a human player against a Minimax AI player with a depth of 4:

python main.py --first-player human --second-player minimax --second-minimax-depth 4

Start a game between a Minimax AI player with a default depth of 6 and a standard random AI player:

python main.py --first-player minimax --second-player random

Project Completed in Course 02180 Introduction to Artificial Intelligence - Technical University of Denmark

chinese-checkers-ai's Issues

PyGame Window does not arrange the matrix as an actual diamond

Solved in PR #30

Fix the utility function to return the correct value based on the terminal test favouring of a player

The UI doesn't keep track of which "pawn" is currently selected

Benchmark the results

1. Count the number of states generated
2. Measure the time of runtime
3. Use different depths for min-max
4. Use different heuristics (Weights, distinct\single heuristic)
5. *Running in cloud
6. *Try comparing without the cutoff function

Somebody should start the logic in the game controller to implement the game loop and switching turns in requesting the next move for each player

Proper ending screen

Within the game logic when the valid actions are generated the logic should be revised to be consistent with the rules

For example after a Step.END we are allowed to do any type of movement except Step.END. After Step.Crawl we should switch turns with the other player and then allow any type of movement except Step.END. After Step.JUMP we should allow any type of movement except Step.CRAWL.

Implement the Minimax algorithm with Alpha-Beta pruning feature within the Minimax AI player

The highlighting effects are displayed always for all the pegs of the player, while I would expect some of them to disappear when the starting peg is selected

Prevent Pull Requests from being merged while tests are failing

Terminal test should be implemented to take into account if the corner is full such that the opponent is stuck in his own corner

During the Minimax execution there is an issue finalising the game as the AI doesn't want to give out when it knows it's losing

The UI does not support displaying possible moves

A random pick from the list of actions can be implement to have a "Confused AI" to play against