Gesture Translation for Heterogeneous Robots

In this paper, we propose a method for translating human gestures to those of various robots that have different degrees and arrangements of freedom. A key idea of the method is to simplify human gesture by using relevant movements of two body parts while making a gesture and explicitly reducing the gesture´s degrees of freedom by utilizing principle component analysis. Experiments were conducted with three types of different types of robots to confirm that this method could produce similar motions to human gestures. The results show that subjects had the impression that the produced motions appeared similar to the motions that humans make with the hand