Impact-Resistant External Torque Estimation for Robot Joint Using Optimized Compliant Model

External torque estimation is important for collab-orative robots. But sharp acceleration or deceleration motion could result in motor current impact, which would lead to undesired estimation error. In this paper, to eliminate the large estimation error caused by the motor current impact, we propose a compliant robot joint model based external torque estimator, and optimize the model through minimizing the weighed sum of maximum and normalized mean squared estimation error. The experimental results on a collaborative robot joint show that the proposed method can effectively estimate the external torque, and resist the error produced by the current impact.