wehub-resource-sync
db1d565b64
2.3 KiB
Step 3: Connect to Your Robot
Now that your AI client has the MCP server configured and rosbridge is running on the robot, you're ready to connect.
1. Connect
Open your AI client and tell it to connect to the robot:
Connect to the robot at <robot-ip>
Replace <robot-ip> with your robot's IP address on the local network (e.g., 192.168.1.42). If the MCP server and ROS are on the same machine, use localhost.
The MCP server will report that the IP is reachable and the rosbridge port is open — this means you're connected.
Make sure the rosbridge port (default 9090) is not blocked by a firewall on the robot's machine.
2. Explore
Once connected, try asking your AI client to explore the ROS system:
What topics and services are available on the robot?
What nodes are currently running?
The MCP server will query rosbridge and return the results from the robot's ROS environment.
3. Try It Out
You can interact with the robot using natural language:
Make the robot move forward
Subscribe to the /odom topic and show me the latest message
If you don't have a physical robot, turtlesim is the standard "hello world" for ROS and is a great option to explore and experiment. Launch it using:
ROS 2:
ros2 run turtlesim turtlesim_node
ROS 1:
rosrun turtlesim turtlesim_node
For a full walkthrough, see the Turtlesim Tutorial.
More Examples
This repo includes several examples to try with different robots and setups:
- Turtlesim — the "hello world" of ROS
- Turtlesim with Docker — no ROS install required
- LIMO Mobile Robot
- Unitree Go2
- TurtleBot3
- Image Topics
For more advanced demos with simulated robots in Gazebo, see the ROS-MCP Demos repository which includes a warehouse TugBot, Unitree Go2 quadruped, and drone control with PX4.