rummicubei

This is an AI robot that can play the game of Rummikub. The game is made in Java. The robot consists of 2 parts Monte Carlo tree search that can find and play the best move possible and a prediction of opponents tiles made by ml to solve the problem of impartial information.

Done by group 8 DACS Maastricht 2023/2024:
-Frederik Grüneberg
-Jakub Suszwedyk
-Kees van den Eijnden
-Kasper van der Horst
-Kaloyan Kostov
-Konrad Paszyński
-Julius Verschoof

Prerequisites

java
maven

Python prerequisites

Python version: 3.10.4

Install the required packages with pip install -r requirements.txt. Pytorch has to be installed separately as it is dependent on the system. For more info refer to the documentation.

How to run

Make sure Python is set up and Python prerequisites are installed properly

JAR file

Download the jar file and launch it you will be trowed in a game selection menu where you can play against other people on the same device or more importantly, if you choose single-player mode you will start a game vs our AI and you can try and beat it

Build from source

You can also download the whole source code and run it from your favorite editor. rummicubei/src/main/java/com/gameEngine/GameEngine.java is the file you need to run to achieve the same result as the jar file

Bot selection

In the main method of the game you can change the bot from baseline to mcts or mcts+ml beware that mcts take some time to generate a move