This project contains the following

• a simple 2D game engine, based around the PyGame library.
• Q-Learning system for adding reinforcement learning to games.
• several sample games, with some reinforcement learning examples, and Tensorflow models, demonstrating the game AI.
• the main learning source code originally evolved from the book Hands-On Machine Learning with Scikit-Learn and TensorFlow, which was modified for TensorFlow 2.

Tensorflow 2, Keras, Python 3.8

Using the requirements.txt file, use

conda create --name <env_name> --file requirements.txt

python launcher.py --game <PATH_TO_GAME_CONFIG.yml>

python launcher.py --game ./games/othello/resources/game4.yml

A spaceship can move horizontally to dodge the falling asteroids. This is perhaps the simplest demo game to develop a Q-Learning based AI.

Othello game (also called Reversi). Tricky to develop Q-Learning based AI for. Work in progress.

┌ proton - main source code for the game engine
|
├ tests - a few tests (not many!)
|
├ games - games
    ├ asteroiddodge - source code for asteroid dodge game
    |    ├ resources - contains the TF models, Learing parameter files, image files etc
    | 
    ├ othello - source code for othello game
    |    ├ resources - contains the TF models, Learing parameter files, image files etc
    |
    ├ piratedefense - source code for a terrible demo game