Abstract
Learning by showing is one of effective methods when robots acquire human skill of actions. Especially if plural sample of human same actions are given, it will be more advantageous to get the skills. To develop such a skill acquisition approach by plural showing, in this paper, as a first step, we present a method to find a time point of a trajectory on human action corresponding to each time point of the other trajectories. In addition, we apply our algorithm to the case of a Japanese kendama action to verify the effectiveness of our approach.
