Robotic skills learning based on dynamical movement primitives using a wearable device

Abstract

It is an effective way for the robots to learn operation skills from the humans. In this paper, we realize a skill learning system based on a teleportation system for transferring the human experience to the robot. Firstly, the robotic teleoperation system with a wearable device is developed by controlling the motor speed directly. This system greatly reduces the time delay by comparing with the way that controlling with point position. Then, a rotation invariant dynamical movement primitive method is presented for learning the operation skills. Finally, the effectiveness of the proposed human experience learning system is evaluated by experiments on a Baxter robot. The human-robot interaction experiment shows the validity of the presented robotic skill learning system.