Sensor-Free Method with BP Network to Achieve Drag Teaching on the 7-DoF Collaborative Robot

Abstract

Drag teaching is an effective way to realize the user-friendly characteristics of the collaborative robot. Aiming at the limitations of traditional drag teaching technology, we propose a teaching method without the external force/torque sensor combined with a neural network. Specifically, the dynamics model of the robot is established based on Lagrange equation, and then the parameters of the dynamic model are identified by a neural network; Based on the servo motor’s current value and combined with the dynamic model, the external force disturbance is estimated equivalently in the drag process; In the stage of trajectory reproduction, spline interpolation is used to smooth the trajectory waypoints. Finally, we verify the effectiveness of the proposed method through simulation and physical experiments.