Robot Motion Support with Sense of Agency for Users

Abstract

　What is the appropriate method for human-robot interactions to support the motions? Sense of agency, that is the subjective awareness of controlling one’s own volitional actions, is easily lost when motions are directory supported by the robot. We have proposed the bio-mimetic behavior adaptation algorithm called tacit learning. Tacit learning can tune the roughly defined behaviors to the sophisticated ones through body-environment interactions. Sophisticated behaviors here mean that the behaviors are more adapted to the environment. We apply tacit learning for controlling the human motion support robots and succeeded to support motions maximizing sense of agency of the users. In this paper, we show the basic idea of human motion support with tacit learning introducing the motion support by the exoskeleton robot and the forearm prosthesis.