scorelesspine / reinforcement_learning_project

AIM240 Capstone Project Part 2 Code Submission

Reinforcement_Learning_Project

AIM240 Capstone Project Part 2 Code Submission

Installation Instructions

Install Python v3.10.4: https://www.python.org/downloads/release/python-3104/

Clone this repository. If using Git, you may run

git clone https://github.com/ScorelessPine/Reinforcement_Learning_Project && cd Reinforcement_Learning_Project

Otherwise download the Zip of this repository from https://github.com/ScorelessPine/Reinforcement_Learning_Project/archive/refs/heads/main.zip and unzip into a directory.

Optional: Create a python virtual environment

python -m venv RL
. RL/Scripts/activate

Install prerequisites

Note: Box2d is not officially supported on Python 3.10, so a package must be manually installed for it to work. Navigate to https://www.lfd.uci.edu/~gohlke/pythonlibs/ Search for "Box2D-2.3.2-cp310-cp310-win_amd64.whl" Download it and place it inside the repository.

pip install setuptools==66
pip install stable-baselines3[extra]
pip install "Box2D-2.3.2-cp310-cp310-win_amd64.whl"
pip install gym[box2d]
pip install pyglet==1.5.27

(NOTE that pip install gym[box2d] will likely error on python 3.10.4, however it installs some requirements for gym box2d that are needed) (NOTE this pyglet=1.5.27 is needed to render the CarRacing environment)

Run the evaluation and testing of the models

python EvaluateAndTestModels.py

Capstone Video

Shown here is a video of the project being evaluated per the AI Capstone Project: https://www.youtube.com/watch?v=LPrLolCQXbs