MCP + LangGraph Integration for ROS 2 Robot Agent

This project integrates the Model Context Protocol (MCP) tool with LangGraph to build an LLM-powered ROS 2 robot agent.

This project utilizes few tools defined in the ROSA repository, as MCP tools.

Features

• Python-based client-server architecture
• Uses MCP to connect local language models
• Integrates with ROS 2 (Humble)
• Designed for robotic decision-making workflows

Files

• client.py: Connects to MCP-enabled server and handles language model interaction, it has beed tested with the LLM model LLaMA 3.1:70B.
• new_server.py: Handles server-side logic and MCP node management

Requirements

• Python 3.10+
• ROS 2 Humble

pip install fastmcp
pip install langchain-mcp-adapters

Usage

Make sure you have ROS 2 Humble installed. You can follow the official guide here: ROS 2 Installation Instructions

🐢 Install and Run Turtlesim

sudo apt update
sudo apt install ros-humble-turtlesim

Run turtlesim in a terminal:

ros2 run turtlesim turtlesim_node

---

🤖 Run ROS Agent (Client)

Clone this repo:

git clone git@github.com:sahars93/Langgraph-mcp-ROS2.git

And, run:

python3 client.py

Demo