An Emotional Support Robot Framework Using Emotion Recognition as Nonverbal Communication for Human-Robot Co-adaptation

Abstract

Human emotion is an essential nonverbal communication tool. Granting machines this type of ability will improve our communication with technology significantly, thus giving us a more natural experience while interacting with machines. Software systems should have the ability to adapt to such nonverbal cues. The focus of our research is the incorporation of human emotions in co-adaptive software systems. Specifically, how emotionally aware systems should react to human emotions. One of the numerous application areas for this promising technology is affective robotics. In this paper, we propose a Framework for a co-adaptive Emotional Support Robot. This framework adopts facial expression recognition as the main method for detecting emotions. In addition, this human-centric framework has a strong emphasis on the personalization of user experience. We adopt a personalized emotion recognition approach, as not all humans show emotions in the same way. As well as the personalization of the system’s adaptive reactions, based on a reinforced learning approach where the system assesses its own actions.