Quick and reliable contact detection for sensorless force control of industrial robots for human support

In this paper, we propose a new contact detection method for sensorless force control of industrial robots for human support. Sensorless force control is an useful technique with regard to force control of industrial robots. However, existing force estimation methods have some error due to the phase-lag of the disturbance observer, and due to the errors in robot dynamics identification. The proposed method achieves quick and reliable contact detection by employing a jerk signal with the dynamic threshold. A jerk signal, which is sensitive to impulsive force and the effect of errors due to dynamics parameters is very small. The proposed dynamic threshold is adjusted according to robot's motion. When the robot exhibits accelerating motion, the threshold is set to a higher level in comparison with the usual low-level constant threshold. The validity of the proposed method is confirmed through the experiments on actual industrial robots.