Realsense Ros Mcp

realsense-ros-mcp

[](https://www.python.org/downloads/) [](https://docs.ros.org/en/humble/) [](https://opensource.org/licenses/MIT) [](https://www.intelrealsense.com/)

🌐 English | [中文](./README.zh.md)

Intel RealSense ROS2 MCP Server — wraps the realsense2_camera ROS2 interface, providing 25 MCP tools for AI assistants to control Intel RealSense depth cameras via ROS2.

---

Features

ROS2 Node Management

• launch_camera — Launch realsense2_camera ROS2 node
• stop_camera — Stop camera node
• list_active_cameras — List active camera processes

Topic Operations

• list_topics — List ROS2 topics
• get_topic_info — Get topic type and publishers
• get_topic_hz — Measure topic publishing frequency
• capture_color_image — Capture color image from topic
• capture_depth_image — Capture depth image from topic
• capture_rgbd — Capture aligned RGBD image pair
• capture_infrared — Capture infrared image
• get_imu_data — Get IMU data
• capture_pointcloud — Capture point cloud (PLY/PCD)

Service Calls

• get_device_info — Call DeviceInfo service

Parameter Control

• get_camera_parameters — Get all parameters
• set_camera_parameter — Set parameter
• enable_filter — Enable/disable filters
• set_depth_profile — Set depth stream resolution and FPS
• set_color_profile — Set color stream resolution and FPS

TF and Calibration

• get_camera_info — Get CameraInfo (intrinsics)
• get_tf — Query TF transform
• get_extrinsics — Get extrinsics

Multi-Camera

• launch_multi_cameras — Batch launch multiple cameras
• capture_multi_frames — Multi-camera synchronized capture

ROS2 Diagnostics

• check_ros2_status — Check ROS2 environment
• list_ros2_nodes — List all nodes

---

Quick Start

Prerequisites

# Install ROS2 Humble (Ubuntu 22.04)
# See: https://docs.ros.org/en/humble/Installation.html

# Install realsense2_camera package
sudo apt install ros-humble-realsense2-camera

# Or install from source
# cd ~/ros2_ws/src
# git clone https://github.com/IntelRealSense/realsense-ros.git -b ros2-development
# cd ..
# rosdep install -i --from-path src --rosdistro humble -y
# colcon build

Installation

# Ensure ROS2 is sourced
source /opt/ros/humble/setup.bash

# Install Python dependencies
pip install -r requirements.txt

# Test server
python mcp_server.py --test

# Run server (stdio)
python mcp_server.py

# Demo (without MCP)
python tests/demo_ros_capture.py

MCP Configuration

Add to OpenClaw config.yaml:

mcp:
  servers:
    realsense-ros:
      command: bash
      args: ["-c", "source /opt/ros/humble/setup.bash && python3 /path/to/realsense-ros-mcp/mcp_server.py"]

Or use Claude Desktop configuration:

{
  "mcpServers": {
    "realsense-ros": {
      "command": "bash",
      "args": ["-c", "source /opt/ros/humble/setup.bash && python3 /path/to/realsense-ros-mcp/mcp_server.py"]
    }
  }
}

---

Hardware Requirements

• Intel RealSense D400/L500 series cameras
• D435/D435i (recommended)
• D455
• D415
• L515
• ROS2 Humble (Ubuntu 22.04)
• realsense2_camera package installed
• USB 3.0 connection (required)
• Python 3.8+

---

Usage Examples

# Check ROS2 status
check_ros2_status()

# Launch camera node
launch_camera(serial="12345", namespace="camera")

# List topics
list_topics()
# Returns: ["/camera/color/image_raw", "/camera/depth/image_rect_raw", ...]

# Capture color image
capture_color_image(topic="/camera/color/image_raw", filename="color.png")

# Get camera parameters
get_camera_parameters(node_namespace="camera")

# Set parameter
set_camera_parameter(node_namespace="camera", param_name="depth_module.emitter_enabled", param_value=True)

# Stop camera
stop_camera(namespace="camera")

---

Architecture

┌─────────────────────────────────────────────────────────┐
│                     MCP Client                          │
│              (Claude / OpenClaw / etc.)                 │
└────────────────────┬────────────────────────────────────┘
                     │ stdio
                     ▼
┌─────────────────────────────────────────────────────────┐
│                realsense-ros-mcp                        │
│              (mcp_server.py / FastMCP)                  │
└────────────────────┬────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────┐
│                   bridge.py                             │
│        (ROS2Bridge - rclpy / ros2cli integration)       │
└────────────────────┬────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────┐
│              ROS2 Humble + realsense2_camera            │
│              (ROS2 nodes / topics / services)           │
└────────────────────┬────────────────────────────────────┘
                     │ USB 3.0
                     ▼
┌─────────────────────────────────────────────────────────┐
│             Intel RealSense Camera                      │
│                (D435/D455/L515/etc.)                    │
└─────────────────────────────────────────────────────────┘

---

Comparison with librealsense-mcp

| Feature | librealsense-mcp | realsense-ros-mcp | |------|------------------|-------------------| | Low-level | pyrealsense2 direct calls | ROS2 topics/services | | Dependencies | pyrealsense2 | ROS2 Humble + realsense2_camera | | Use Case | Standalone apps, lightweight | ROS2 robotics integration | | TF Support | ❌ | ✅ | | Point Cloud Format | PLY | PLY / PCD | | Filters | SDK built-in | ROS2 dynamic parameter config | | Multi-camera | Manual management | ROS2 native support |

---

Testing

# Ensure ROS2 environment is set
source /opt/ros/humble/setup.bash

# Run tests
cd tests
python demo_ros_capture.py

---

Related Projects

• librealsense-mcp — Direct SDK version of RealSense MCP Server
• realsense-ros — Official ROS2 RealSense package

---

License

MIT License - See [LICENSE](LICENSE)

---

Made with ❤️ for the ROS2 robotics community

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: ros-claw
• Source: ros-claw/realsense-ros-mcp
• License: MIT

Install and usage instructions live in the source repository linked above.