Tele-impedance control of virtual system with visual feedback to verify adaptation of unstable dynamics during reach-to-point tasks

摘要

Over the past few decades, robot actuators with variable stiffness or impedance control have been developed. This control adjusts the stiffness of the task space of joint space, enabling the robot to perform explosive tasks and to interact with environments and humans. However, planning the robot stiffness is a difficult task, except in a few optimal control problems, such as maximizing the output speed or distance, which can be solved analytically. We propose that humans can program robot stiffness by a technique called tele-impedance control, which transfers the stiffness and equilibrium points of electromyogram and position sensors from a human operator, respectively. However, tele-impedance control allows only visual feedback between a robot and an operator. Therefore, we first verify whether a human operator can adapt the stiffness under unstable dynamics during a reach-to-point task. In two types of experiments, the operator adapted to the unstable dynamics and generated a stiffness trajectory that effectively and efficiently completed the task.