Comprehensive compensation method for position and attitude errors of industrial robot

Abstract

The invention discloses a comprehensive compensation method for position and attitude errors of an industrial robot. The comprehensive compensation method comprises the steps of constructing binocularvision measurement of the tail end actual position and attitude of the robot, and sampling the tail end actual position and attitude errors of the robot under large-scale configuration; and establishing a training data set and a verification data set of the theoretical position and attitude and the corresponding errors of the robot, and achieving comprehensive prediction and compensation of the position and attitude errors of target points of the robot by training a deep learning model. By means of the comprehensive compensation method, comprehensive compensation of the position and attitudeerrors of the robot can be achieved; based on sampling data under the large-scale robot configuration, the distribution rule of the position and attitude errors of the robot can be better reflected; and the position and attitude errors of the industrial robot can be effectively reduced.

Description

technical field [0001] The invention relates to the technical field of position and attitude error compensation of industrial robots, in particular to a comprehensive compensation method for position and attitude errors of industrial robots. Background technique [0002] Industrial robots have the advantages of high flexibility, high efficiency, flexible joints, good openness, and low maintenance costs. They are important components of intelligent manufacturing. The precision of industrial robots is an important factor affecting product quality. However, due to errors caused by the manufacturing, assembly, connecting rod deformation, and gear clearance of industrial robots, the absolute pose accuracy of industrial robots is much lower than the repetitive pose accuracy. The low absolute pose accuracy of industrial robots will lead to a sharp decline in product processing quality. Especially in advanced manufacturing fields such as aircraft assembly combined with off-line pro...

AI Technical Summary

Problems solved by technology

Patent CN110385720A discloses a robot positioning error compensation method based on a deep neural network, constructs a neural network model, takes the theoretical pose of the robot and the corresponding position error as training samples, and the trained neural network model can predict the position error of the robot target point , this method does not need to build a complex motion error model, and at the same time considers the influence of the robot's positioning attitude on the positioning error, but the interpretability of the error prediction process is weak, and the prediction and compensation of the robot's attitude error cannot be realized

The existing robot accuracy compensation methods generally have the following shortcomings: only the position error compensation of the robot target point is realized, and the attitude error is not compensated; the sampling points of different robot configurations are too small to fully reflect the position error and attitude of the robot The distribution of errors

Images