Improving task skill transfer method by acquiring impedance parameters from human demonstration

This paper addresses how to acquire reusable robot skills for practical tasks based on a programming by demonstration approach. Specifically, this paper aims at improving the task skill transfer method, which has been proposed by the authors, by acquiring impedance parameters and equilibrium trajectories simultaneously from demonstration data. First, the overview of the task skill acquisition method is described. Then, how to acquire the impedance parameters is discussed. Finally, the effectiveness of the proposed method is verified by experiments.