CITS3001 Mario Project

Alex Hawking (https://github.com/Alex-Hawking) & Connor Grayden

Description

This project investigates the performance of Deep RL models on the game Super Mario Bros through the package gym-super-mario-bros. The two algorithms used were PPO (Alex) and DDQN (Connor), so both options are available for training and testing.

Installation and Usage

To install this project, follow these steps (each model will have its own directory holding code, these steps need to be executed within the directories):

DDQN

Follow these steps to get up and running:

1. Create a conda environment using mario.yml:

   conda env create --name DDQN_Mario --file mario.yml

2. Activate the environment

   activate DDQN_Mario

3. Ensure Correct Environment Setup

   ├── DDQN
   │ ├── DDQN_Model.pt
   │ ├── DDQNtrain.pt
   │ ├── DDQNrun.py

4. Run
   To train the model:

   python DDQNtrain.py

   To run the trained model:

   python DDQNrun.py

PPO

Follow these steps to get up and running:

1. Create a virtual environment:

   python3 -m venv venv

   (Would highly recommend)

2. Activate the virtual environment:

   • Linux/Mac:

     source venv/bin/activate

   • Windows:

     .\venv\Scripts\activate

3. Install the required packages:

   pip install -r requirements.txt

4. Run

   python3 main.py