https://github.com/alirezamika/bipedal-es
AI learning to walk in gym's BipedalWalker environment.
https://github.com/alirezamika/bipedal-es
Last synced: about 1 year ago
JSON representation
AI learning to walk in gym's BipedalWalker environment.
- Host: GitHub
- URL: https://github.com/alirezamika/bipedal-es
- Owner: alirezamika
- Created: 2017-06-29T06:15:24.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2017-06-29T20:01:40.000Z (almost 9 years ago)
- Last Synced: 2025-03-28T00:41:26.160Z (about 1 year ago)
- Language: Python
- Size: 20.5 KB
- Stars: 66
- Watchers: 4
- Forks: 13
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# An AI agent learning to walk in gym's BipedalWalker environment.
This AI agent uses Evolution Strategies and deep learning models to learn how to walk.
Read [Evolution Strategies as a Scalable Alternative to Reinforcement Learning](https://blog.openai.com/evolution-strategies/) from OpenAI if you are interested.
After a few hundred iterations, he can walk! Or he may run ;)
# Dependencies
- [evostra](https://github.com/alirezamika/evostra)
- [gym](https://github.com/openai/gym)
# Usage
To see the agent walking:
```
from bipedal import *
agent = Agent()
# the pre-trained weights are saved into 'weights.pkl' which you can use.
agent.load('weights.pkl')
# play one episode
agent.play(1)
```
To start training the agent:
```
# train for 100 iterations
agent.train(100)
```