Impact Force Reduction Strategies To Achieve Safer Human-Robot Collisions

Abstract

The increasing use of robots operating close to people has made human-robot collisions more likely. In this paper, strategies intended to reduce the impact force to a safe level, without sacrificing the robot’s performance, are investigated. The strategies can be applied to a robot arm without modifying its internal hardware. They include the existing strategies: lowering the actuator controller’s stiffness; actuator switched off upon impact detection; withdrawing the arm upon impact detection; and adding a compliant cover. We also propose the novel strategy of limiting the controller’s feedback term. The collision scenario studied is a robot arm colliding with a person’s constrained head. An improved lumped parameter model of the constrained impact is proposed. Simulation results are included for a UR5 collaborative robot. Sixteen combinations of the impact force reduction strategies are compared. The results show that using a high stiffness controller with a feedback limit and compliant cover reduces the impact force to a safe level, and achieves precise trajectory tracking.