Quantifying Human Mental States in Physical Human-Robot Interaction

Abstract

To support independent living among older adults, we propose the introduction of robots as domestic service providers capable of assisting with daily activities like showering, toileting, and transferring. These tasks often require physical interaction, necessitating robots with efficient physical capabilities and a nuanced social understanding of humans, such as stress levels and attitudes toward the robot (robot perception). Our objective is to explore potential real-time (objective) quantification approaches in physical Human-Robot Interaction (pHRI), with the ultimate aim of developing a human mental state predictor to enhance robots’ decision-making abilities. We identify cognitive ergonomics and belief as two key human factors associated with cognitive aspects that can sufficiently represent the human mental state. To accurately measure these dimensions, we recognize the need for a combination of objective and subjective approaches. However, literature examining the suitability of subjective measures in pHRI, which is crucial for obtaining reliable objective quantification, is limited. Therefore, our research includes an assessment of a critical subjective quantification tool for a key human factor, i.e., robot perception, followed by an exploration of correlations between subjective and objective measures. Our findings reveal significant correlations between subjective and objective measures, indicating the potential for objective indicators such as skin temperature and task performance to reliably infer cognitive ergonomics and belief in collaborative pHRI scenarios. Furthermore, we observe promising associations between pre-gathered data, such as personality traits and demographics, and measurable dimensions of cognitive ergonomics and belief, suggesting a supplementary approach for enhancing inference using objective measures.