Affective State Estimation and Robot-aided Physiological Monitoring of Workers in Hazardous Environments

This PhD thesis delves into the intricate realm of worker's health monitoring, encompassing unobtrusive human-centred approaches, each contributing to the overarching goal of advancing the health and safety of the worker in occupational environments. In Industry 5.0, milestones include human-centered workplaces and worker health monitoring, which benefit both individual employees and the entire production facility. Attention to the health monitoring of workers results in the prevention of health risks and proactively taking preventive measures. Despite global efforts, occupational accidents and work-related diseases persist as major concerns, with millions of lives lost annually. Wearable and contactless devices can be advantageously applied in various working scenarios, including improving the quality of human-robot interaction in remote robotic teleoperations, monitoring workers' health in occupational environments, and detecting vital parameters of survivors in search and rescue operations. The objective of this thesis is to propose novel health monitoring methods, leveraging machine learning for affective state estimation in human-machine interaction and developing robot-aided monitoring methods for workers' vital parameters in hazardous environments, including both regular and emergency situations. Proposed monitoring methods of the workers' health state are validated through experiments involving healthy subjects in the real or simulated working environment. The research presented in this thesis has shown to provide a preliminary significant positive impact on workers' safety in several applications, including human-robot interaction, complex work environments exposed to high risks, and search and rescue operations, leading to emerging potential future perspectives on worker well-being and overall efficiency