Research On Position And Pose Measurement Of Shaft Hole In Auto-assembly System Based On Vision

With the continuous development of science and technology,traditional manufacturing has been unable to meet the needs of customized users.Faced with the transformation and upgrading of traditional manufacturing,intelligent manufacturing has attracted people s attention.Intelligent assembly is made possible by the assembly robots with vision,but the current research in the field of intelligent assembly cannot meet the requirements of the intelligent assembly industry.In order to further improve the positioning accuracy and intelligent level of assembly robots,taking typical shaft-hole assembly as the research object,this paper focuses on the calibration of ETH(Eye-to-Hand)vision system and the determination of robotic grasping position and pose,the position and pose measurement of partially occluded objects based on the platform binocular vision system,and the position and pose measurement of hole based on EIH(Eye-in-Hand)vision system.The specific work is as follows:1.The calibration of ETH vision system and the determination of robotic grasping position and pose.Firstly,based on the calibration principle of ETH vision system,the camera calibration and hand-eye calibration are completed by Zhang's calibration method and Tsai-Lenz s algorithm respectively.Secondly,image processing such as median filtering,thresholding,sub-pixel contour extraction and dual-bus extraction is carried out on the workpiece shaft,and the image-based position and pose of the workpiece shaft is determined by the pixel coordinates of the four endpoints of the dual-bus.Finally,the coordinates transformation and hand-eye calibration results are used to realize the transformation of the position and pose of workpiece shaft image to the basic coordinate system of the robot.The experimental results show that the maximum error of the translation part of the hand-eye calibration is 0.669 mm and the maximum error of the rotating part is 0.418°.2.The position and pose measurement of partially occluded objects based on the platform binocular vision system.The proposed method is as follows:Firstly,the left and right images of the object template are respectively matched with the right and left images of the partially occluded obj ect to establish two affine models;Secondly,the pixel coordinates of the images of the occluded obj ect are estimated by using the pixel coordinates of the key points of the template according to the corresponding model;Based on the principle of binocular stereo vision reconstruction,the spatial coordinates of some coplanar key points on the occluded obj ect are solved.Finally,the position and pose is determined by using the solved spatial points.The experiments show that the proposed method is feasible and effective.3.The position and pose measurement of hole based on EIH vision system.Firstly,hand-eye calibration and camera calibration are carried out based on the principle of EIH vision system calibration.Secondly,hole's center of two hole images captured in parallel directions by hand-eye camera are extracted by Hough transform.Aiming at the problem of inaccurate matching of hole's edge points,a method of estimating the basic matrix by SURF+RANSAC matching is proposed,and then the polar lines are used to constrain matching hole's edge points.The horizontal binocular vision model is established.According to the disparity principle,the three-dimensional coordinates of the center and edge points of the hole are solved,and the pose of the hole is solved by space vector method according to the three-dimensional coordinates.Finally,the transformation from the position and pose of the target hole to the basic coordinates of the robot is realized by using the result of EIH hand-eye calibration.The experimental results show that the maximum translation error and rotation error of hand-eye calibration are 0.792mm and 0.2210°respectively,and the maximum position error of the robot is 2.44mm.4.Accuracy verification of binocular vision measurement.The chessboard calibration board is fixed at the end of the robot,and the actual relative rotation angles between the calibration boards are obtained by controlling the rotation of one joint of the robot.The sizes of the chessboards and the angles between the normal vectors of the calibration boards are measured by binocular stereovision and compared with the actual values.The experimental results show that the measurement accuracy of binocular vision is high.