Eliciting Empathetic Expression Through Robot-Initiated Touch in Companion Scenarios
Abstract
Companion robots are becoming increasingly integrated into everyday life, making the ability to express empathy a critical aspect of their design. However, prior research has focused primarily on verbal and visual cues, leaving robot-initiated touch largely underexplored. To address this gap, we first conducted a pilot study to identify six representative robots whose proactive touch was most acceptable across three morphology types (Human-like, Animal-like, and Machine-like). We then conducted an elicitation study to examine users’ expectations for robot-initiated touch across four emotional states (Happy, Depressed, Tense, and Relaxed). The results showed that both emotional states and robot morphology shaped preferred touch locations and touch behaviors, yielding a set of 24 user-preferred touch gestures. Finally, a Wizard-of-Oz study in a virtual environment demonstrated that a balanced subset of these gestures effectively conveyed empathic intent and improved both users’ affective states and their perceptions of the robots. These findings provide design implications for robot-initiated touch in future companion robots.