Video Game AI Pathfinding Experiment - Q-Learning, Deep Reinforcement Learning using Stable Baselines 3 and A* Search - Artificial Intelligence Foundations - MSc. Computer Science - September 2022/January 2023
This experiment was developed as a means of finding the most effective method of video game pathfinding between three different (informed search) methods. The methods used consisted of Q-Learning, Deep Reinforcement Learning using Stable Baselines 3 and A Star Search. All three methods were written and tested within a Google Colab environment in Python. Each method is also written in separate Python files alongside the main Jupyter Notebook file to allow for local testing if you wish.
It is recommended to test run the Jupyter Notebook file in a Google Colab environment before testing and optimising performance on your own local machine.