A Phased Optimization Method for Robot Dragging Teaching without Torque Sensors

Aiming at the requirements of the robot dragging teaching in practical application, the dragging process is divided and the corresponding dragging control strategy is adopted, so as to improve the flexibility of the robot dragging. On the basis of the known robot dynamics model, the disturbance observer based on generalized momentum is used to estimate the external force of the robot, and the admittance control algorithm is adopt to realize the dragging following the external force. In order to further improve the performance of the dragging teaching, the whole dragging process is divided into static state, traction state and braking state, and the corresponding control strategy is proposed to optimize the starting stage and braking stage in the dragging process. Experimental results show that: increasing torque compensation can effectively reduce the starting time of robot joints, and the starting time can be reduced by 40.22% ~ 51.31% compared without torque compensation; the introduction of variable stiffness parameters in the braking stage can effectively reduce the braking time of robot joints, and the braking time can be reduced by 77.65% ~ 83.97% compared with no braking scheme.

I. Introduction

Human-robot collaboration is a hot research topic in the field of robotics. And teaching function, as an important part of human-robot collaboration, is developing towards intelligence and openness. At present, teaching function of robot is mainly realized by teaching devices or offline programming. The former is simple and intuitive to use, but the process is cumbersome, and the work efficiency will decrease sharply when the operation task needs multiple teaching points; the latter has the advantages of strong commonality and high accuracy, but operators need have some skills and operational experience. The method that the operator manually guides the robot in real-time is called dragging teaching, which can solve the above problems, and improve the simplicity and efficiency of robot teaching operation [ 1 – 2 ].