Sensorless Collision Detection and Contact Force Estimation for Collaborative Robots Based on Torque Observer

Collaborative robots share the same workspace with humans, and it is inevitable that the robot will contact with humans or environments, so the importance of robot collision detection and end-effector contact force estimation has been receiving much attention. The conventional collision detection methods are based on extra sensors, which are not only increase cost but also increase the complexity of robot control system. This paper presented collision detection and contact force estimation methods without any extra sensors. Firstly, we designed a collision torque observer based on robot dynamics model and generalized momentum. Secondly, the joint friction parameters were identified using a friction torque model in the harmonic gear drive system. Finally, the system model errors were eliminated by observing the collision torque under the theoretical condition. The proposed algorithm achieved good results after considering friction factors and system model errors. The results of simulation on MATLAB and ADAMS verified that the algorithm is valid and reliable.