Industrial robot zero point calibration method based on dynamic learning

Abstract

The invention relates to an industrial robot zero point calibration method based on dynamic learning. The method comprises the following steps that a plurality of posture data of an industrial robot are acquired by using a simple calibration point end, and the joint angles of postures and the XYZ distance of a tool coordinate system are recorded; a tail end position expression of each posture is obtained through the corresponding joint angle and the XYZ distance, and the minimum distance error under the plurality of postures is obtained according to the tail end position expressions; the partial differential of the minimum distance error to the joint angle deviation and the XYZ deviation is solved; the above-mentioned two steps are repeated, the partial differential is iterated according to the dynamic learning rate until the minimum distance error is smaller than 1 mm; and the joint angle deviation and the XYZ deviation are compensated to the robot to complete zero point calibration.According to the method, the simple calibration point end is used for collecting the plurality of posture data of the industrial robot, the dynamic learning algorithm is used for carrying out rapid iterative processing on the data, so that the calibration time is controlled to be 30 minutes or within, and the calibration precision is 1 mm or within.

Description

technical field [0001] The invention relates to the technical field of calibration methods for industrial robots, in particular to a zero-point calibration method for industrial robots based on dynamic learning. Background technique [0002] The zero point is the benchmark for the industrial robot to judge its own position. Before the robot leaves the factory, the manufacturer will manually set the zero point of the robot according to the zero mark. However, due to the differences in the robot processing process, the real zero position of each robot is not the same, so it is necessary to perform accurate zero point calibration on the robot. At present, the main zero point calibration methods include laser tracker calibration method, manufacturer's special auxiliary tool calibration method, etc. Such techniques have the following disadvantages: [0003] (1) The required devices are expensive and the calibration process is cumbersome. Taking the laser tracker as an example...

AI Technical Summary

Problems solved by technology

[0005] The present invention proposes a zero-point calibration method for industrial robots based on dynamic learning. A simple calibration tip is used to collect multiple attitude data of industrial robots, and a dynamic learning algorithm is used to quickly iteratively process the data, so that the calibration time is controlled within 30 minutes. Moreover, the calibration accuracy is within 1mm, which solves the shortcomings of the existing calibration method, such as expensive equipment and complicated calibration process

Images