https://github.com/jihoonerd/rl-maze

Simple maze solver by reinforcement learning
https://github.com/jihoonerd/rl-maze

maze policy-gradient q-learning reinforce reinforcement-learning sarsa tutorial

Last synced: 2 months ago
JSON representation

Simple maze solver by reinforcement learning

• Host: GitHub
• URL: https://github.com/jihoonerd/rl-maze
• Owner: jihoonerd
• Created: 2019-05-19T03:29:12.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2024-04-27T15:28:38.000Z (9 months ago)
• Last Synced: 2024-05-22T00:05:08.653Z (8 months ago)
• Topics: maze, policy-gradient, q-learning, reinforce, reinforcement-learning, sarsa, tutorial
• Language: Python
• Homepage:
• Size: 1.54 MB
• Stars: 5
• Watchers: 4
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# rl-maze

## Description

Repository for learning basic concept of reinforcement learning.

This provides following approaches to solve maze.

* Random walk

* Policy gradient (REINFORCE)

* Sarsa

* Q-learning

## How to use

* Instantiate Q-learning agent and train:

```python

agent = Agent('q') # should be one of ['randwalk', 'pg', 'sarsa', 'q']

agent.train()

```

* Draw agent's move by using `agent.state_history`:

```python

maze = Maze()

maze.save_animation('maze.gif', agent.state_history) # it requires imagemagick

```