Robot WebSocket Server

A FastAPI WebSocket server for interacting with Raspberry Pi robots. Supports face recognition, LLM-based conversation, and handles multiple robot clients simultaneously.

Features

• WebSocket-based communication
• Face recognition using face_recognition library
• LLM integration (OpenAI API, with mock mode for testing)
• Async support for multiple robot clients
• Comprehensive test suite with mock clients

Setup

1. Create a virtual environment and install dependencies:

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install -r requirements.txt

2. (Optional) Set up environment variables in .env:

   OPENAI_API_KEY=your_api_key_here
   KNOWN_FACES_DIR=path/to/faces/directory
   

   If these are not set, the server will use mock responses.

Usage

1. Start the server:

   python app.py

   Server runs on ws://localhost:8000/ws/robot

2. Run mock tests:

   pytest tests/test_mock_client.py -v

API Reference

Robot → Server Messages

{
    "text": "Optional text from speech",
    "frame": "Optional base64 encoded JPEG frame",
    "status": "Optional robot status info"
}

Server → Robot Responses

Success:

{
    "speech": "Text for the robot to speak"
}

Error:

{
    "error": "Error message"
}

Project Structure

• app.py - FastAPI application and WebSocket endpoints
• schemas.py - Pydantic models for message validation
• vision.py - Face recognition functionality
• llm.py - LLM integration and response generation
• tests/test_mock_client.py - Mock client tests