Universal Robots MCP server

A Model Context Protocol (MCP) server for Universal Robots, enabling AI agents and applications to control and monitor UR robots through a standardized interface.

Overview

This project provides a comprehensive MCP server implementation that exposes Universal Robots functionality through a standardized API. It allows AI assistants, automation tools, and other MCP-compatible clients to interact with UR robots for monitoring, control, and program management.

This is a modern usage example of the UnderAutomation.UniversalRobots library, demonstrating how to integrate Universal Robots capabilities with the Model Context Protocol framework to create AI-powered robot control solutions.

Features

Robot Control

• Power Management: Power on/off the robot and control brake release
• Program Execution: Play, pause, and stop robot programs
• Program Management: Load programs and installations
• Variable Access: Get and monitor robot variables

Dashboard Functions

• User Interface Control: Show and close popup messages on the teach pendant
• Status Monitoring: Check if robot is in remote control mode, if programs are running
• System Information: Retrieve Polyscope version, robot model, and loaded programs

Real-Time Data Access

Access comprehensive real-time robot data through the Primary Interface:

• Robot Mode Data: Current operational mode and status
• Joint Data: Individual joint positions, velocities, and currents
• Cartesian Info: TCP position and orientation in Cartesian space
• Tool Data: Tool communication and mode information
• Force Mode Data: Force control parameters when in force mode
• Safety Data: Safety system status and parameters
• Configuration Data: Robot configuration and kinematics info
• Program Threads: Active program execution threads
• Messages: Runtime exceptions, popup messages, text messages, and key messages

File System Access

• Program Enumeration: List all available programs on the robot
• Installation Enumeration: List all available installations
• SFTP access for file management

Technology Stack

• .NET 9.0: Built on the latest .NET platform
• ASP.NET Core: Web application framework for HTTP transport
• ModelContextProtocol: MCP server and ASP.NET Core integration (v0.4.0-preview.3)
• UnderAutomation.UniversalRobots: Universal Robots SDK (v8.2.0)

Prerequisites

• .NET 9.0 SDK
• Universal Robots robot (physical or simulator)
• Network connectivity to the robot

Configuration

The robot connection is configured in Program.cs:

var parameters = new ConnectParameters("192.168.56.101");
parameters.Dashboard.Enable = true;
parameters.Rtde.Enable = false;
parameters.PrimaryInterface.Enable = true;
parameters.Ssh.EnableSftp = true;
parameters.Ssh.Username = "ur"; // use "root" for real robot, "ur" for simulator
parameters.Ssh.Password = "easybot"; // default password

Update the IP address and credentials according to your robot setup.

Usage

1. Build the project:

   dotnet build

2. Run the MCP server:

   dotnet run

3. Connect your MCP client to the server endpoint to access all available tools.

Available MCP Tools

The server exposes over 30 tools including:

• power_on / power_off - Robot power management
• play / pause / stop - Program execution control
• load_program / load_installation - Load robot programs and installations
• show_popup / close_popup - Teach pendant UI control
• get_robot_model / get_polyscope_version - System information
• get_joint_data / get_cartesian_info - Real-time position data
• get_safety_data / get_force_mode_data - Safety and force control status
• enumerate_programs / enumerate_installations - File system browsing
• And many more for comprehensive robot monitoring and control

Use Cases

• AI-Assisted Robot Programming: Enable AI agents to control and monitor robots
• Automated Testing: Integrate robot control into automated test workflows
• Remote Monitoring: Access real-time robot data through standardized APIs
• Multi-Robot Orchestration: Control multiple robots through a unified interface
• Educational Applications: Teach robotics concepts with AI-powered assistance