Effects of Human Emotions, Robot Reliability, and Team Hierarchy on Human-Robot Interaction

Abstract

Social robots gradually become a vital part of human lives as they can assist humans in complex tasks. To enhance the effectiveness and user acceptance of human-robot interaction (HRI), it is critical to understand what factors could improve human trust in robots. This dissertation builds on an existing framework of human-related, robot-related, and environmental factors in human-robot trust. Emotions, a human-related factor, have been discovered to have critical impacts on HRI, but research has focused more on robots' emotional expressions than user emotions. To bridge this research gap, three human-centered studies were conducted in the present dissertation investigating the impact of users' emotions (happiness and anger) along with robot reliability (robot-related) and team hierarchy (environmental factor) on their trust toward robots in a collaborative setting, an escape room, with a social robot, Pepper. Additionally, user perceptions, perceived workload, and task performance were also measured in the dissertation. The experimental setting and procedures for the three studies were similar. To induce participants' emotions, they were asked to write down their emotional experiences regarding the assigned emotions (anger vs. happiness) for 12 minutes, then solve puzzles in an escape room with Pepper. An emotional manipulation check was completed by the participants by filling out an emotional evaluation survey three times: before the induction, after the induction, and after the completion of the escape room. In addition, user perceptions, trust, perceived workload, and task performance were also collected through validated questionnaires and observations. Study I focused only on the effect of human emotions, anger and happiness (N = 46), suggesting that happy participants rated the robot agent as significantly more likable, safer, and more comfortable, while angry participants complied significantly but fewer succeeded. Among those who failed, angry participants showed higher cognitive trust in the robot. Study II (N = 86) explored emotion and robot reliability (high vs. low) with a between-subjects design. Happy participants rated the robot as more anthropomorphized, intelligent, and likable, with higher trust. Participants in the high-reliability condition perceived better performance, complied more, and responded faster than the participants in the low-reliability condition. Study III (N = 52) investigated emotion and team hierarchy (human-follower-robot-leader vs. human-leader-robot-follower) with a mixed-subjects design. Participants in the follower role perceived the robot as less responsive but complied more. Angry participants perceived the robot as more capable of constructive responses than happy participants while happy participants showed significantly more verbal interactions with the robot. An interaction between emotion and hierarchy was shown for social presence and perceived performance. Interviews highlighted the role of emotions and hierarchy in user perception, decision-making, emotional changes, and responsibility for team success or failure. In conclusion, the results underscored the importance of user emotions, robot reliability, and team hierarchy in shaping user perceptions and trust in robots and task performance, providing valuable theoretical and practical implications for emotions in HRI.