Differential torque deviation collision detection method based on the dynamic model of industrial robot

Aiming at the problem of industrial robot collision detection, a differential torque deviation collision detection method based on the dynamic model of industrial robots is proposed. An adaptive genetic algorithm based on penalty function is used to calculate the optimal excitation trajectory in the identification of dynamic parameters. A single-layer neural network with an adaptive learning rate is used to identify the dynamic parameters of the robot, and the robot dynamic model is obtained. The torque deviation between the predicted torque and the actual torque is calculated by using the dynamic model, the sensor carried by the robot and the driving current signal.The backward first-order difference and backward second-order difference of the torque deviation are carried out to distinguish the industrial robot collision into intentional collision and unintentional collision, realize the robot collision detection, and trigger the corresponding collision avoidance mechanism to avoid equipment and personnel damage. Physical collision detection is verified with 6R industrial robots, and the experimental results show that the collision detection algorithm can effectively distinguish non-collision, unintentional collision and intentional collision, which improves the safety of collaborative operation of industrial robots.