LLM-Driven FMEA for Safe Human-Robot Collaboration in Disassembly

Abstract

Disassembly operations often present unstructured and unpredictable scenarios, such as handling hazardous materials, addressing ergonomic strain, and managing dynamic robot interactions that pose safety risks. To tackle these challenges, we propose an innovative use of large language models (LLMs) to enhance failure mode and effect analysis (FMEA) in the context of human-robot collaboration (HRC) for disassembly tasks. We developed an LLM system leveraging retrieval-augmented generation (RAG) for real-time risk analysis and recommendation generation. RAG retrieves domain-specific information from the FMEA knowledge database, enabling accurate risk analysis, contextual understanding, and relevant recommendations based on user input and operational data. Evaluation of the model demonstrated precision, achieving a BLEU score of 86.5, a ROUGE-L score of 89.2, and a cosine similarity score of 0.92 compared to ISO reference documents. These results underscore the system's capability to produce domain-specific recommendations closely aligned with safety technical specifications. The system accelerates the FMEA process by identifying failure modes, prioritizing risks, and proposing mitigation strategies in real time, enhancing adaptability to collaborative disassembly complexities. This model proactively addresses hazards like toxic material exposure or robotic arm glitches, transforming risk management with precision and efficiency. A customized prototype demonstrates its effectiveness, featuring real-time risk analysis, domain-specific knowledge retrieval, and context-aware recommendations. Its user-friendly design and adaptability to dynamic scenarios, such as hazardous material handling, showcase its potential to redefine risk assessment in collaborative settings.