Robot collision detection based on dynamic model

Abstract

Aiming at the security problem of cooperative robots, this paper presents a low cost collision detection algorithm without torque sensors. In this method, the collision can be detected with the sampling torque and the dynamic model. First, the relationship between the output torque and the motion state of the rotating joint of the robot is established according to the Newton Euler method. For the unknown parameters in the dynamic model, experiment and identification are adopted. In order to reduce the influence of sampling noise and error on the parameters, the excitation trajectory is designed according to err analysis and the identification of the dynamics parameters is calculated with the least squares method. In the actual operating environment, the collision detection is performed according to the torque deviation and setting threshold between the sampling torque and the dynamic model torque. Finally, the experiment is carried out on the body of ER3A of Effort. The experimental results show that this method can detect the occurrence of collision sensitively.