D5.3 Human robot collaboration behavioural analysis and modelling

This deliverable analyses human AI collaboration within the AI-PRISM project, focusing on how AI-driven automation supports manufacturing work while preserving operator safety, skills, and job quality. Across multiple industrial use cases, baseline analyses shows that manual tasks expose operators to significant physical, cognitive, and psychosocial demands, often relying on tacit, experience-based skills. Participatory design and task analysis demonstrate that these risks can be reduced through targeted automation that complements rather than replaces human work. Operators consistently value systems that remove physically demanding, repetitive, or error-prone activities while maintaining human oversight, decision-making, and quality judgement. Derived human-centred design requirements emphasise transparency, control, adaptability, and meaningful human involvement. Validation of AI-PRISM demonstrators indicate reductions in workload and physical strain, improve usability and trust, and particular benefits for novice operators. Overall, the findings confirm that human-centred design, appropriate human–AI function allocation, and integrated workforce training are essential to achieving safe, acceptable, and sustainable human AI collaboration in manufacturing.