ros2

This MCP server integrates FastMCP with ROS 2 to allow AI assistants to control a robot by publishing Twist messages to the /cmd_vel topic. It runs as a ROS 2 node and accepts movement commands from MCP clients (such as Claude) via FastMCP's asyncio pipeline. The server translates high-level commands like move forward at a specified speed for a given duration into ROS 2 Twist messages, enabling time-based control of a real or simulated robot. You can verify behavior by monitoring the /cmd_vel topic and observing the robot's response to issued commands.

Prerequisites:\n- ROS 2 Humble installed and sourced.\n- Python 3.10.\n- uv (Python package manager) for dependency management.\n- numpy and fastmcp Python packages.\n\nInstallation steps:\n1. Clone the repository:\nbash\ngit clone https://github.com/kakimochi/ros2-mcp-server.git\cd ros2-mcp-server\n\n2. Use the Python version specified (3.10) via a local environment as shown in the repository. A .python-version file is provided:\nbash\ncat .python-version\n3.10\n\n3. Create and activate a uv virtual environment:\nbash\nuv venv --python /usr/bin/python3.10\nsource .venv/bin/activate\n\n4. Install Python dependencies in editable mode:\nbash\nuv pip install -e .\n\n5. Ensure ROS 2 Humble is sourced in your shell and that ROS is available to Python.\nbash\nsource /opt/ros/humble/setup.bash\n

Environment variables and ROS setup issues are common. If you encounter ROS logging directory errors, set ROS_LOG_DIR to a writable directory (e.g., /tmp). Confirm you are using Python 3.10 for ROS 2 Humble compatibility. The MCP server currently supports only the /cmd_vel topic with Twist messages; extending to other topics would require additional code changes. When configuring clients like Claude, ensure the path to the repository is correctly mounted in the uv command.