yuva-i / ai-term-project-final-submission

DSCI 6612 - AI - Term-Project - Final-Submission

Contributors

Yuva Krishna Kishore Inapala, Rahul Gaddam

Introduction

In this project we have applied reinforcement learning and trained a smart agent which is going to take action based on enemies actions and takes a counter action.

Requirements

Python 3.6 to 3.8, OS - Ubuntu, Optuna, Jupyter Notebook

Install dependencies:

pip install -r requirements.txt

Game

Teenage Mutant Ninja Turtles - Tournament Fighters

Game Setup

For this project we have used stable-retro librar, which is a fork of of gym-retro ('lets you turn classic video games into Gymnasium environments for reinforcement learning') with additional games, emulators and supported platforms. Along to it we have used openCV to monitor the actions and 'stable baselines3' for modeling and training using PPO(Proximal Policy Optimization) model.

Methods we used

We have rescaled the resolution of the game output and grayscaled the screen for faster and easy processing on local machine.

Evaluation and Metrics

We can check average reward at the end of each episode, enemy health, and score using info function and can get performance visualization for each model while it is learning using tensorboard.

User Feedback

We welcome your feedback! If you encounter issues or have suggestions, please create an issue.

License

This project is licensed under the MIT License.