OpenRacer

Overview

The OpenRacer WebSocket interface facilitates communication between the Unity-based simulator and Python-based machine learning models. The simulator sends input data for each frame, and the ML model processes it before sending a response back.

This package provides a server interface that manages WebSocket connections and ensures seamless real-time data exchange.

Installation

pip install OpenRacer

Usage

Creating a Custom Model

To integrate an ML model, create a class that inherits from ModelBase in the OpenRacer package and implement following methonds

• preProcess (optional) : used for processing input before passing to trainEval or testEval
• trainEval : will be called each frame for training
• testEval: will be called each frame for testing/Race/eval
• rewardFn: will be called in each eval to monitor model performance
• backprop : will be called for each step in training. Use this to update model based on eval.

from OpenRacer.Model import ModelBase

class MyModel(ModelBase):
    # implement the required functions 

Starting the Interface

The Interface is responsible for handling connections between the simulator and the ML model and Dashboard.

from OpenRacer.Interface import Interface

# Create and start the server
Interface(model=modelInterface).start()

API Reference

Will be updated in some time

Notes

• Ensure that the WebSocket server is running before launching the simulator.

License

This package is open-source under the MIT License.

Contributing

Contributions are welcome! Feel free to submit issues or pull requests on GitHub.

Contact

For questions, reach out to the OpenRacer team via GitHub.