Multi-Robot Control MCP Agent

This project creates an MCP (Model Context Protocol) server that allows AI agents to control multiple robots via existing FastAPI robot control servers.

Features

• Control multiple robots simultaneously
• Control robot movement (forward, backward, left, right)
• Play sounds through the robots' speakers
• Get robot status information
• Precise velocity control
• Access robot camera images
• Unified API with port specification

Prerequisites

• Python 3.10+
• UV (Python package manager)
• Multiple running robot control FastAPI servers (as provided in the example)

Setup

1. Ensure you have a Python 3.10 environment active

2. Install dependencies using UV:

uv pip install -e .

Usage

1. Start the robot control FastAPI servers:

2. Important Update: The Claude desktop client now automatically runs the MCP server for you

3. The MCP server allows AI agents to control multiple robots and access their cameras.

Available Tools

All tools accept a `port` parameter (default: 8000) to specify which robot to control.

Movement Control

• `drive_forward`: Move a robot forward
• `drive_backward`: Move a robot backward
• `turn_left`: Turn a robot left
• `turn_right`: Turn a robot right
• `stop`: Stop robot movement
• `drive`: Control with precise velocity values

Audio Control

• `beep`: Play a sound through a robot's speaker

Camera Access

• `get_camera_image`: Get an image from a robot's camera

System Information

• `robot_status`: Get robot status information
• `list_available_robots`: List all available robots and their status

Available Resources

• `robot://info/{port}`: Get information about a specific robot's capabilities

Examples

# Get status from robot on port 8000
status_robot1 = await client.robot_status(port=8000)

# Get status from robot on port 8001
status_robot2 = await client.robot_status(port=8001)

# Make both robots beep with different tones
await client.beep(port=8000, frequency=440, duration=1.0)  # A4 note on robot 1
await client.beep(port=8001, frequency=523.25, duration=1.0)  # C5 note on robot 2

# Get a list of all available robots
robots = await client.list_available_robots()

Note on Image Handling

The camera image tools use MCP's native `Image` class for handling image data. This allows the AI agent to receive the image data in a format that can be properly handled by the client without need for additional conversion.