Structured light hand-eye calibration method based on coordinate system servo alignment

Abstract

The invention relates to the field of three-dimensional hand-eye calibration, and aims at three-dimensional calibration of a structured light camera coordinate system and a robot tail end coordinate system. A structured light sensor is installed at the tail end of a robot, and the three-dimensional relative pose relation of a structured light camera coordinate system is estimated with high precision through an off-line and on-line two-step estimation method by means of geometric information of a specially-designed calibration block. The method specifically comprises the steps that by means of special geometric information of the calibration block, an operator can manually adjust the posture of the tail end of a robot, a camera is driven to move to collect images, and the initial relative pose of the structured light camera coordinate system is rapidly obtained; and after the initial relative pose is obtained, the robot can drive the camera to collect images through automatic operation including a specified path, the relative pose of the structured light camera coordinate system is estimated in an algorithm loop iteration mode, and finally a high-precision calibration result is obtained. The method has high practicability in robot hand-eye vision application based on structured light, such as welding seam tracking and robot automatic polishing.

Description

Technical field [0001] The present invention belongs to the field of computer visual, and specifically, the present invention relates to a structural optical camera coordinate system and a hand-eye scale method for the end coordinate system of the robot. Background technique [0002] With the development of science and technology and the continuous improvement of industrial automation, the robots have been widely used in various industries. The combination of visual and robots makes the robot a pair of "eyes", structural light as a proactive visual ratio compared to traditional 2D plane vision, robustness, can also obtain three-dimensional data; in the weld tracking or robot automatic grinding project, the combination of structural light and robot has received a lot of applications. Therefore, the hand-eye calibration is an indispensable link in the hand-eye visual application of the structural optical robot. The traditional three-dimensional hand-operated method often causes the...

AI Technical Summary

Problems solved by technology

The traditional 3D hand-eye calibration method often uses a robot to drive the camera to move, collect images under multiple poses on the calibration board, obtain enough calibration data, and then complete the hand-eye calibration through matrix fitting, pose decomposition and other steps; this method has many disadvantages: (1) The attitude decomposition process often faces the problem of insufficient accuracy; (2) When there is a slight disturbance in the system, the result consistency of the fitted matrix after attitude decomposition is poor; (3) During the calibration process, the choice of the robot’s photographing attitude It plays a vital role in the calibration results, and the calibration accuracy is overly dependent on the professional level of the operator;

Images