F1Tenth @ UCI_race_stack

A Race Car Simulation

powered by DreamerV3

Get Started · Documentation · Report Bug

Table of Contents

1. Introduction
   • Built With
2. Features
3. Contributing
4. Testing
5. Checklist
6. License

Introduction

We add DreamerV3 implementations of World Models into the F1tenth ROS2 environment in PyTorch.

Built With

We referenced the following repositories for the base code:

• dreamerv3-torch

• racing_dreamer

• dreamerv3

Features

This project:

• Bridges the gap between 1D LiDAR data and DreamerV3's input requirements.

• Includes a custom development environment (Gymnasium).

• Leverages multi-core training environment to maximize computational efficiency.

• Provides seamless integration with the ROS2 framework.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please reference CONTRIBUTING.md for detailed contribution guide.

Testing

• implemented f1tenth gym for testing environment.
• waiting on issac sim implementation for better simulation environment.

Checklist

• basic structure implementation(bare bones)
• LiDAR feed for encoded latent state
• World Model implementation
• Actor-Critic implementation
• Scaling reward system

License

This project is licensed under the MIT license.