Motion retargeting for humanoid robots based on identification to preserve and reproduce human motion features

This paper presents the method to retarget human motion. The method can evaluate the ability of the preservation of the original characteristics of human motion data. It enables to compute the joint trajectories of the human corresponding with the retargeted ones of the robot at the same time, by utilizing the geometric identification technique. The obtained trajectories of a human are the solution to minimize the cost function about motion reproduction. The proposed method is efficient for such applications that the robot needs to mimic human motion without modifying the detailed features of the original movement of each body segment. The results of the retargeted motions to a humanoid robot are shown.