Enabling Modular and Intelligent IoRT Systems : Integrating the Model Context Protocol for Semantic Decoupling at the Edge

Abstract

The Internet of Robotic Things (IoRT) is shifting the paradigmatic of device-environment engagement through the seamless converge of sensing, actuation, and intelligent reasoning. However, contemporary IoRT systems exhibit strong inter-dependencies between devices and application services, which hinders extensibility and prolongs technological innovation cycles, especially in the context of the rapidly evolving Edge AI landscape. In this paper, we propose a novel integration of the recently-introduced Model Context Protocol (MCP) into edge-level IoRT systems to enable modular service composition and semantic decoupling. By treating devices as callable resources and tools, MCP allows large language models (LLMs) to orchestrate complex workflows through natural language prompts. We design a unified framework where heterogeneous IoRT nodes expose their capabilities through MCP-compliant interfaces, enabling scalable interaction and task automation. A proof-of-concept implementation based on digital twin edge nodes demonstrates the feasibility of the approach, including modular access and decoupled service. We also discuss key challenges such as semantic standardization, security, and edge-level optimization. This work opens new avenues for building generalizable, intelligent, and adaptable IoRT platforms.