Jingcui01 / AIVoid

Neural network reinforcement Q-learning for an avoidance game

AIVoid

This is a simple game in python which can be played by both human player and an AI Player. AI has been trained via neural network reinforcement learning. See YouTube video

Usage:

Run RunGame.py to start playing, select the options and Human vs AI player (AI pre train vs AI post train choice at game run selects from 'dumb' pre-train and 'smart' post-train AI). To train select the learning and game parameters in GameParameters and LearnParameters in Training.py (see descriptions in Game and Bots modules and the report) and run Training.py. Training can be done for a fixed number of training sessions or until the test score achieves a certain threshold.

Dependencies:

• Python 2.7
• Numpy python package
• Tkinter python package

Authors:

Tonci Antunovic