Building of Empathy Inducing mModel between Robot and Human based on the Affect Expressing Ontology - experimental study on applicability of empathy inducing non-verbal language of human to robot

Abstract

In the age of digital technology, it would be preferable for an agent to possess the capability to induce empathy from human, for it to be used effectively and efficiently. We tried to identify the empathy inducing non-verbal languages from human through the experiment to introduce them to an agent. We built the environment to measure the extent of empathy and observe expression from groups of subjects, using questionnaire, eye tracker and video camera. Two subjects were installed at once, one named observee and the other observer. One with eye tracker was observer and asked to infer the result of impression evaluation made by observee on meaningless stimuli shown on a display. Inference of observee's impression evaluation made by observer and observee's real result are compared and used as an index to see the extent of empathy. Utilizing this index, we specified the moment/stimuli in which observer empathized with obsevee. As a result of this experiment, we found several non-verbal languages, from the groups of stimuli in which observer empathized, through analysis on video record of subjects. They, however, failed to be supported by recorded data of eye tracker.