Skip to content
Onitama Board Game Simulator with Reinforcement Learning opponents
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


This project was made for CS 5033 Machine Learning. It emulates the Onitama board game.

The associated paper with motivation, methods, and results can be viewed here.

Run Instructions

Use Python 3.6 or higher

python3 -m pip install -r requirements.txt

Set Experiments and Agents

Place new agents in src/agents. See others as example. BaseAgent will act as a human player, but is also the base class for all agents

Place new experiments in src/experiments. See others as example. Make sure to set blue_agent, red_agent, and do_render. game_ended() will be called after every round with the final game state. Agents can be reassigned at any time (e.g. swapping blue_agent and red_agent every match)

Tell PyOnitama which experiment to use in as the argument to the Game constructor. (Line 10 at time of writing)

You can’t perform that action at this time.