Iterative Learning-based Trajectory Generation of Robot Manipulator to Reproduce Force Response of Teaching Device

Abstract

This study proposes a method for correcting trajectory data of a robot using iterative learning to mimic the motion of a human with a comparable force level. Currently, trajectory generation using direct teaching is often used to teach robots more flexible movements. Although such a teaching method can produce ideal trajectories, it may neglect the reproducibility of the force. Therefore, we propose a method that can reproduce the originally required sense of force by correcting the original trajectory data using iterative learning. Moreover, the reduction in the force error using our method is verified.