Compensation method and system for friction torque of robot joints

Abstract

The invention discloses a robot joint friction torque compensation method and system. Through the compensation method and system, in the industrial robot operating process, after such parameters as joint temperature, load, lubrication and wear are changed, a friction model can be correspondingly changed, so that the failure problem of a feedforward compensation method based on a fixed friction model can be prevented. On the basis, the identified friction model is used for predicting a generated friction torque training neural network, so that the training learning process efficiency of the neural network is higher, the ideal effect can be quickly achieved, and the robot joint tracking speed and tracking precision are improved. Further, the friction torque feedforward compensation is performed through Gaussian sampling to make up the defect of no exploration in neutral network outputting, so that the neural network can preferably restrain to prevent local optimum, the tracking precisionis further guaranteed, and even if such parameters as joint temperature, load, lubrication and wear are changed, the purpose of high-precision tracking can be quickly achieved.

Description

technical field [0001] The invention relates to the field of robots, in particular to a method and system for compensating friction torque of robot joints. Background technique [0002] Joint friction of industrial robots is one of the main factors that reduce the tracking accuracy of robot joints. In order to reduce the impact of friction on the robot control system, the friction torque must be compensated. Usually, the friction torque compensation adopts the feed-forward compensation method based on the fixed friction model, but due to the constant changes in the parameters such as joint temperature, load, lubrication and wear during the operation of the industrial robot, the friction model of the feed-forward compensation changes significantly, thus This leads to the failure of the feed-forward compensation method based on the fixed friction model, which reduces the tracking accuracy of the robot joints. Contents of the invention [0003] The purpose of the present inv...

AI Technical Summary

Problems solved by technology

Usually, the friction torque compensation adopts the feed-forward compensation method based on the fixed friction model, but due to the constant changes in the parameters such as joint temperature, load, lubrication and wear during the operation of the industrial robot, the friction model of the feed-forward compensation changes significantly, thus This leads to the failure of the feed-forward compensation method based on the fixed friction model, which reduces the tracking accuracy of the robot joints

Images