Skip to content

Unity Environments based on ML-agents for the Walker agent to perform various locomotion behaviors using reinforcement learning

Notifications You must be signed in to change notification settings

TroddenSpade/Walker-Motion

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Walker Motion

This repository includes several Unity Environments based on ML-agents for the Walker agent to perform different motion scenarios using reinforcement learning. Each environment comes with its own set of challenges and rewards and is designed to challenge the Walker agent to learn a specific locomotion behavior, including Toe Walking, Knee Walking, Target Chasing, Hopping, Werewolf Running, and Getting Up.

Getting Started

To get started with this repository, you will need to have Unity and the ML-agents toolkit installed on your machine. Once installed, you can clone the repository to your local machine using the following command:

git clone https://github.com/troddenspade/Walker-Motion.git

Then, open the Unity project files and explore the different environments. You can train your Walker agent by running the training scripts provided for each environment.

Scenarios

Toe Walking

In this environment, the Walker agent must learn to walk on its toes without falling over.

video_2023-04-10_02-24-43.mp4

Knee Walking

In this environment, the Walker agent must learn to walk on its knees without falling over.

video_2023-04-10_02-24-42.mp4

Target Chasing

In this environment, the Walker agent must learn to follow a target placed randomly within the field without falling over.

video_2023-05-03_12-53-56.mp4

Hopping

In this environment, the Walker agent must learn to hop on one leg.

video_2023-05-04_13-46-57.mp4

Werewolf Running

In this environment, the Walker agent must learn to run like a werewolf on hands and feet.

video_2023-05-05_23-30-13.mp4

Getting Up (+ Intrinsic Reward)

In this environment, the Walker agent must learn to get up from a lying position.

video_2023-05-11_20-21-38.mp4

Kip Up

video_2023-05-11_23-44-54.mp4

Getting Up (No Exploration)

video_2023-05-11_23-44-55.mp4

About

Unity Environments based on ML-agents for the Walker agent to perform various locomotion behaviors using reinforcement learning

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published