SETTE: A Human-Centric Framework for Operator Safety Monitoring in Industry 5.0

Manufacturing environments in Industry 5.0 demand adaptive, human-centric safety systems that can dynamically respond to evolving regulations and operational conditions. Despite technological advances, human error accounts for 70-90% of industrial accidents, highlighting the need for intelligent safety monitoring systems. We present SETTE - Safety Environment using Tracking & ioT for Enterprises - a novel hybrid framework that integrates Large Language Models, Extended Reality, and Internet of Things technologies for comprehensive operator safety monitoring. SETTE operates through two interconnected phases: design time, where multi-agent LLM architectures automatically extract safety rules from heterogeneous regulatory documents (e.g., ISO 45001, company manuals) and encode them in a Domain-Specific Language with embedded Extended Reality simulation metadata; and run time, where validated rules are enforced through real-time monitoring using IoT sensors, computer vision, and wearables integrated with an AI-powered Safety Management System. A key innovation is the human-in-the-loop validation process, where safety experts interact with automatically generated Extended Reality scenarios to assess rule feasibility before deployment, ensuring both regulatory compliance and practical applicability. The framework maintains full traceability between generated rules and source regulations while providing adaptive, context-aware guidance through a conversational Human-Machine Interface. We illustrate the framework’s operational logic through a running example involving machine guarding rules for rotating machinery, showing how regulatory text can be systematically transformed into executable safety rules with complete audit trails. Unlike traditional static safety protocols, SETTE enables continuous adaptation to changing regulations and operational conditions, providing a transparent safety layer that enhances worker understanding and compliance while reducing human error in manufacturing environments.