Webots MCP Controller

This project provides a framework for controlling a NAO robot in the Webots simulator using the Model-Context-Protocol (MCP). It allows for interaction with the robot using natural language through compatible clients.

• File-Based Communication: A robust communication system between the MCP server and the Webots controller using json files.
• Motor Position Validation: All motor commands are validated against the robot's physical limits to prevent errors.
• Motion Playback: Play pre-recorded .motion files.
• Sensing: Access to camera and GPS data.
• Actuator Control: Control head, arms, and LEDs.
• Dynamic Robot Discovery: The server can handle multiple robots, identifying them by their ROBOT_NAME.

The system is composed of two main components:

1. webots_mcp_server.py:

   • This script runs the FastMCP server, exposing the robot's capabilities as tools.
   • It communicates with the Webots controller by writing commands to a commands.json file and reading the robot's state from a status.json file.
   • It handles multiple robots by creating separate data directories for each.

2. controllers/my_controller_plus_mcp/my_controller_plus_mcp.py:

   • This is the main controller script for the NAO robot in Webots.
   • It runs in the Webots simulation environment and is responsible for all low-level robot control.
   • It continuously reads the commands.json file for new commands from the MCP server.
   • It writes the robot's current status (motor positions, sensor data) to the status.json file.

1. Open the project in Webots: Launch Webots and open the worlds/MCP-test.wbt world file.
2. Start the simulation: Click the "Play" button (▶).
3. Connect a client: Connect to the project using an MCP client (e.g., Gemini CLI). The client is responsible for automatically starting the webots_mcp_server.py script based on its configuration. For example, a settings.json file for the client might contain:

   "I-robot-mcp": {
     "command": "python.exe",
     "args": [
       "d:/webots-mcp/webots_mcp_server.py"
     ]
   }

4. Control the Robot: Once the client is connected and the server is running, you can start interacting with the robot.

The server provides the following tools for controlling the robot:

• get_visual_perception(robot_name): Get an image from the robot's camera.
• get_robot_position(robot_name): Get the robot's current position.
• get_robot_status(robot_name): Get the full status of the robot.
• set_head_position(robot_name, yaw, pitch): Set the head position.
• set_arm_position(robot_name, arm, shoulder_pitch, shoulder_roll): Set the arm position.
• reset_robot_pose(robot_name): Reset the robot's pose.
• list_motions(): List all available motions.
• play_motion(robot_name, motion_name): Play a pre-recorded motion.
• set_led_color(robot_name, color, part): Set the color of the robot's LEDs.
• get_robot_capabilities(robot_name): Get information about the robot's capabilities.
• check_webots_connection(robot_name): Check the connection with the Webots controller.
• list_robots(): List all active robots.