Generation of various hand-waving motion of a humanoid robot in a greeting situation

Abstract

To enhance the quality of human-robot interaction, the ability of a robot to make human-like expressions is regarded as being a key issue. Although there are several expression channels, we have focused on expression that is based on gestures. While humans can generate appropriate gestures easily and unconsciously, it is very difficult for a robot to do so. In this study, we focus on those gestures that rely on repeated motions, as in the case of waving one's hand in a greeting situation involving two people. Since a person's hand can be waved any number of times in such a situation, trajectory-based gesture learning is not well suited to learning the motions that have the same meaning as the gesture. Therefore, we suggest the use of space-based gesture learning as a solution for learning repeated motions having the same meaning and demonstrate the possibility of space-based gesture learning by applying the system to the DARwIn-OP humanoid robot.