A Pose Accuracy Compensation Method for Industrial Robots Based on Deep Learning

Abstract

The invention discloses a method for compensating position and posture accuracy of industrial robots based on deep learning. The invention combines the error similarity of industrial robots and proposes a deep learning model for compensating position and posture errors of industrial robots, including the construction of two deep beliefs The network-parallel model architecture uses the state characteristics of industrial robots and their corresponding pose error parameters as sample data to train the deep learning model by means of combination and comparison, and uses the error similarity of industrial robots as the network model supervision The additional features of learning, the deep learning model after training can more accurately predict and compensate the position and posture errors of industrial robots; the present invention combines the advantages of deep learning feature expression ability and statistical explainability As an additional feature of supervised learning, comparative information other than the robot state feature itself is introduced to improve the prediction accuracy of the deep learning model.

Description

technical field [0001] The invention relates to the technical field of industrial robot pose error compensation, in particular to a method for compensating the pose accuracy of industrial robots based on deep learning. Background technique [0002] With the implementation of the intelligent manufacturing strategy, industrial robots have been gradually used in high-end manufacturing due to their high efficiency, high flexibility, and high degree of automation. However, the absolute pose accuracy of industrial robots is far lower than the repetitive pose accuracy. As a result, industrial robots cannot meet the precision requirements of high-end manufacturing products, especially in the application scenarios of large-scale and complex structures combined with off-line programming. The low absolute pose accuracy has become the main obstacle restricting the application of industrial robots in intelligent manufacturing. Accuracy Compensation Technology [0003] With the advantage...

AI Technical Summary

Problems solved by technology

The existing robot accuracy compensation methods can be roughly summarized as the following deficiencies: only the position error compensation of the robot target pose is realized, and the attitude error is not compensated; the sampling step size and distribution have a great influence on the accuracy of accuracy compensation; error prediction The interpretability of the model is weak

Images