LLM-Driven Multi-Agent Control for Skill-Based Smart Manufacturing

2026-10-01

AI Summary

This study addresses the challenges of frequent reprogramming and the inability of static programs to handle runtime faults in small-batch, highly customized manufacturing. To this end, it proposes a large language model-based multi-agent system that coordinates factory modules via the Model Context Protocol (MCP) and MQTT communication. By integrating OPC UA, the approach synergizes offline sequence generation with online real-time control, establishing a standardized MCP toolchain and a real-time state injection mechanism. Experimental results demonstrate that both monolithic and peer-to-peer architectures achieve an average solution rate of 93%, while the orchestrator architecture effectively handles silent faults, exhibiting emergent diagnostic behavior without explicit logic. These findings validate the feasibility of multi-agent collaboration in intelligent manufacturing.

Abstract

Factories are shifting toward smaller lot sizes with high product customization, requiring frequent re-programming of flexible and reconfigurable automation systems. LLM-based agents can be deployed in two complementary roles: Offline, they generate deterministic production sequences, reducing programming effort; online, they operate live machines and handle unforeseen runtime faults that static programs cannot anticipate. We propose a solution in which each factory module is paired with a dedicated LLM-based agent and an MCP tool server that exposes the module's skills via OPC UA method calls, with agents coordinating over MQTT and grounded by real-time updates of the factory state. We compare three agent architectures (orchestrator, peer-to-peer, and monolithic) across nine production challenges of increasing complexity in a simulation of a physical six-module hexagonal factory, including silent hardware fault detection. The monolithic and peer-to-peer architectures both achieve the highest mean solve rate (93\%), while the orchestrator uniquely resolves a silent conveyor-belt fault in all ten runs by autonomously rerouting plates around the blocked segment. All architectures exhibit emergent fault-diagnosis behavior without any explicit failure-handling logic, establishing standardized MCP tooling, MQTT-based inter-agent communication, and real-time state injection as a viable and reproducible foundation for LLM-programmed smart manufacturing.

Problem

• Smart Manufacturing
• Multi-Agent Control
• Large Language Models
• Fault Diagnosis
• Reconfigurable Automation

Innovation

• LLM-driven multi-agent
• Model Context Protocol (MCP)
• Smart Manufacturing
• Emergent fault diagnosis
• Agent architectures