Raffaella Loffredo
Towards Human-Centric Industry 5.0 Production: A Closed-Loop LLM-RAG System for Operator Wellbeing-Driven Cobot Adaptation.
Rel. Alessandro Simeone, Domenico Augusto Francesco Maisano. Politecnico di Torino, Master of science program in Industrial Production And Technological Innovation Engineering, 2026
Abstract
Operator psychological state in Human-Robot Collaboration (HRC) environments remains systematically excluded from real-time production monitoring: physiological wearable sensors have failed to achieve industrial deployability after two decades of research, and LLM-based manufacturing decision support systems have been uniformly designed around production-side data, treating operator state as a non-variable. This work proposes a closed-loop LLM-RAG Decision Support System that integrates real-time operator wellbeing monitoring, evidence-based diagnostic reasoning, deterministic safety enforcement, and mandatory supervisory control into a single operational architecture. Digital psychometric self-reports (NASA-TLX, VAS-F, SSSQ, SAM, RAS), administered through an adaptive bounded probe loop, replace physiological instrumentation. A RAG module (ChromaDB, all-MiniLM-L6-v2) grounds a GPT-4o-mini diagnostic agent in a curated knowledge base of anonymous operational cases.
The agent constructs situational hypotheses without predefined diagnostic labels, reasoning explicitly across five evidence dimensions
Relators
Academic year
Publication type
Number of Pages
Additional Information
Course of studies
Classe di laurea
Ente in cotutela
Aziende collaboratrici
URI
![]() |
Modify record (reserved for operators) |
