Control Stiffness-Based Motion Reproduction for Environmental Impedance Variation

Motion generation technology contributes to the automation of the robots, especially in the industrial field. Imitation learning and motion reproduction system (MRS) are one of the motion generation methods that widely applied. However, machine learning-based methods require a large amount of data and are weak against extrapolation. Although MRS has merit in terms of the data number, it is weak regarding environmental variation. In particular, the environmental impedance variation have occurred, both position and force tracking are impossible. As several conventional studies have been done to reduce only position or force tracking errors, they can not adapt force or position priority tasks, respectively. Therefore, this paper proposes a novel compensation law based on designed control stiffness using the hybrid angle. The validity of this proposal is confirmed by simulation. This paper contributes to the MRS expansion of the environmental impedance variation with consideration of the task's priority physical value.