Understanding safety performance in human-robot interaction: the role of operator fatigue, emotion, and workload

Purpose The increasing adoption of intelligent robotics in construction has improved operational efficiency while introducing new safety challenges in human-robot interaction (HRI). Although operator-related factors are recognized as important, the ways in which fatigue, emotion, and workload relate to safety performance during collaborative operation remain insufficiently examined. This study aims to investigate these relationships through a controlled experiment to understand how multidimensional operator states systematically impact safety outcomes in HRI. Design/methodology/approach A controlled experiment engaged 31 operators performing simulated ground-leveling tasks under low, medium, and high fatigue conditions. A rigorous statistical framework was established to analyze task efficiency, accuracy, and safety performance. The study utilized Linear Mixed-Effects Models (LMMs) for continuous metrics and Generalized Estimating Equations (GEEs) with a Poisson distribution and a log link for count data to ensure robust statistical inference. Findings Increasing fatigue was associated with longer task completion times and greater trajectory deviation. Performance decline was already observable at medium fatigue levels, indicating a progressive rather than abrupt deterioration pattern. Collisions and out-of-range events rose substantially when transitioning from low to high fatigue conditions. In addition, more negative emotional states and higher perceived workload corresponded to reduced operational efficiency and safety performance. The results suggest that multiple operator-state dimensions may jointly relate to measurable changes in efficiency, spatial control, and safety event occurrence during HRI. Social implications By highlighting the role of operator states in HRI, this study emphasizes the need to integrate human factors into the deployment of intelligent construction systems. Incorporating operator-state awareness into task planning and monitoring may help reduce preventable incidents, improve worker well-being, and support the safer adoption of robotic technologies on construction sites. Originality/value This study examines fatigue, emotion, and perceived workload within a unified experimental framework and relates these operator dimensions to observable safety indicators in HRI. By embedding operator conditions assessment within an operational construction task and linking it to measurable efficiency, spatial accuracy, and safety events, the study provides empirical support for integrating multidimensional operator conditions into safety management practices in intelligent construction environments.