Hand-eye calibration method based on deep learning

Abstract

The invention provides a hand-eye calibration method based on deep learning. The method comprises the following steps: collecting and marking a training image containing a calibration object; training a yolo3 model; setting a moving step length S and a distance threshold value D; enabling a control device to randomly adjust the position posture of the mechanical arm, and enabling a camera to shoot and obtain an adjusted image; and calling the trained yo3 model to analyze the adjustment image, if the size of the calibration object is not smaller than half of the view field of the camera and the actual distance is not larger than the distance threshold value D, shooting the calibration image and recording the position posture data of the mechanical arm at the moment, otherwise, shooting the adjustment image again after adjusting the mechanical arm according to the moving step length S, shooting a calibration image again; and after the image is calibrated to meet the required number, making hand-eye calibration matrix calculation. Manual participation is not needed in the process, manpower and material resources are saved, the calculation amount is small, the operation speed is high, the shooting requirement for the calibration image is low, the result accuracy is high, and robustness is better.

Description

technical field [0001] The invention relates to a hand-eye calibration method based on deep learning. Background technique [0002] With the continuous improvement of the level of industrialization, the application of vision-guided technology in the field of industrial robots is becoming more and more extensive. Through vision-guided industrial robots, the robot can not only perform preset actions, but also locate the position of the grasped target through machine vision. Perform a soft grip. The hand-eye calibration of industrial robots is the basis for realizing machine vision guidance, that is, the conversion matrix between the camera coordinate system and the robot coordinate system is determined through hand-eye calibration. Most of the existing hand-eye calibration methods use manual intervention for data collection, that is, by manually controlling the movement of the robotic arm, the calibration object is located in a suitable position, and the posture of the roboti...

AI Technical Summary

Problems solved by technology

However, there are still some defects in their actual use: the former needs to obtain the center position information of the current calibration object and the image size of the calibration object through corner point calculation. If the field of view of the camera is too small, it will be difficult to extract the corner points, that is, the position information of the calibration object cannot be obtained, and the robustness is poor; the latter solves the cantilever effect of the robotic arm through three-point calibration, and three-point calibration is required before each calibration. Manual intervention by staff with some expertise is still required

Images