Exploring the Role of Heat Exchange During Tactile Human-Robot Interaction

Comprehending human perception of tactile connection with humanoid robots is essential for enhancing human-robot interaction (HRI). In this context, affective haptics, a study domain examining how technologies may impact human emotional states via tactile sensations, is becoming a crucial element of HRI. Importantly, although affective haptics recognizes the role of touch temperature in human interactions, research examining the impact of heat exchange through the touch in HRI remains limited. In this context, infrared thermography (IRT) may serve as an appropriate instrument to examine the impact of robotic touch temperature on HRI without disrupting the ecological characteristics of the interaction. This research investigates how the temperature of robotic touch influences human comfort during the interaction with the NAO robot through IRT. Participants were instructed by the robot to grasp its hands as it recounted a brief story characterized by moderate arousal and positive emotional valence, lasting around 30 seconds. The procedure was conducted under three distinct experimental conditions, modulating the robot hands temperature (i.e., hot, cold, and neutral conditions). The findings indicated a reduced level of comfort experienced in cold condition compared to hot condition (p=0.013). Regarding the IRT analysis, differences were evaluated for skewness and sample entropy of the temperature time course of the nose tip of the subjects (i.e., skewness: neutral vs. cold, p=0.016, and neutral vs. hot, p=0.040; sample entropy: cold vs. hot, p=0.015, and cold vs. neutral, p=0.025). Additionally, a machine learning method achieved a 3-class classification of comfort reported by participants relying on the IRT features with an accuracy of 71.4%. These results demonstrate the significance of accounting for heat exchanges in the development of affective haptic interfaces for robots. To achieve this goal, IRT provides a viable and effective method for monitoring.