Human combination motion recognition based on haptic information using similarly structured master-slave robot hand

In this paper, a motion recognition method that uses haptic information of a human hand is proposed. The haptic information is measured by a master/slave robot hand. This robot has five degrees of freedom, and a bilateral control of the robot is implemented. An operator wears on the master robot hand and operates an object through the slave robot hand. A motion database is prepared to contain haptic information as reference vectors concerning eight kinds of human motion. Then, the motion database utilizes cosine similarity to distinguish the human motion from haptic information acquired by the master robot. Furthermore, the expansion and contraction of time axes are corrected by dynamic programming matching, and combination motion is consequently recognized. The validity of the proposed method is experimentally confirmed.