https://github.com/chloedia/pacman
Creation of a PacMan environment and benchmark of different state of the art agent
https://github.com/chloedia/pacman
pacman q-learning-vs-sarsa reinforcement-learning
Last synced: 6 months ago
JSON representation
Creation of a PacMan environment and benchmark of different state of the art agent
- Host: GitHub
- URL: https://github.com/chloedia/pacman
- Owner: chloedia
- Created: 2022-03-29T17:02:12.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2022-03-29T18:32:50.000Z (over 3 years ago)
- Last Synced: 2025-02-08T20:17:51.488Z (8 months ago)
- Topics: pacman, q-learning-vs-sarsa, reinforcement-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 1.84 MB
- Stars: 1
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Pac-Man 
This is our Project for the CentraleSupelec's Reinforcement Learning course, a Pac-Man. The
purpose of this work was to train in the best way possible different agents (Q-
Learning, Expected-Sarsa, Deep Q-Learning) in a Reinforcement Learning environ-
ment of the Game Pac-Man.
## Authors 🧠
- Chloé Daems
- Anne-Claire Laisney
- Amir Mahmoudi
## Requirements 💻
Python 3.9
- matplotlib
- numpy
- turtle
## Getting Started 🐣
You can see our pipeline in the .ipynb file with the trainings and results of the Q learning and expected Sarsa Agents;
If you want to try our DQN.py, you can run the following command on the terminal :
```
$ python DQN.py
```
## Some Results 💰
Here you can find examples of the turtle render for the small and medium grids for the trained Sarsa Agent.
## Implementation ✍️
You can find the details of agents implementations in our report: 'PacMan_Report.pdf'.