Development And Application Of Robot Vision System Based On Structured Light

3D reconstruction is a high-precision,high-efficiency,non-contact measurement technology that plays an important role in industrial inspection,reverse engineering,additive manufacturing,industrial automation,and other fields.Among them,the 3D machine vision system based on line structured light has significant advantages such as high measurement efficiency,simple structure,strong flexibility,and diverse application scenarios.This article focuses on the development of a robot vision system based on line structured light and carries out key technical research on 3D surface reconstruction.In view of the problems existing in this technology,such as complex algorithms,expensive equipment,and poor industrial applicability,the main research contents of this article are as follows:1)A structure light vision sensing device was constructed,consisting of a CMOS industrial camera and a line laser.A computer software system was developed using the QT and Visual Studio platforms,along with relevant libraries and auxiliary software.This system facilitated research on 3D reconstruction algorithms and visual systems.Additionally,a robot laser cladding system was assembled,comprising a six-axis industrial robot,a laser cladding head,a cooling water device,a powder feeding device,and a protective gas cylinder.Communication between the robot repair system and the visual system was achieved through the use of ABB robot’s Robot Studio offline programming software.This experimental platform constitutes the foundation of the research discussed in this paper.2)The calibration of a 3D reconstruction system was conducted,which included the calibration of the vision sensing device and the hand-eye calibration between the ABB robot and the vision sensing device.The calibration of the vision sensing device involved camera calibration and plane calibration.The camera’s intrinsic and extrinsic parameters were calibrated using Zhang’s calibration method.The plane calibration was achieved by utilizing the projective invariance property.By acquiring image information from camera calibration and plane calibration using transformed target pose,the hand-eye calibration was performed.A linear model for hand-eye calibration was constrained and optimized using the properties of rotation matrix vectors.The least squares algorithm was applied to solve the model,enabling the conversion from object’s 2D image information to the robot’s end-effector base coordinates.The system calibration accuracy was evaluated using the mean absolute error,with calibration errors reaching approximately 0.3mm.3)Point cloud data was generated by extracting the center of laser stripes,enabling the realization of 3D reconstruction.A "two-step" algorithm for stripe center extraction was proposed: firstly,a skeleton thinning method was employed for coarse extraction of the stripe center,and then the grayscale centroid method with bilinear interpolation was used for accurate extraction of the stripe center.The PCL(Point Cloud Library)was utilized for point cloud data preprocessing and surface reconstruction algorithms.Point cloud data was refined through processes such as passthrough filtering,statistical outlier removal,and voxel grid filtering,completing the point cloud preprocessing.The rolling-ball algorithm was employed to achieve3 D reconstruction of the point cloud data.4)In order to address the issues of low efficiency and lack of automation in measuring non-standard components and repairing damaged parts,the experimental platform developed in this study was used to measure and repair surface pits on stainless steel.Experimental measurements were conducted on the diameter and depth of the pits.Subsequently,the point cloud data was processed,and the system software generated an STL model and performed model slicing.Based on the results of the slicing,the process parameters of the robot repair system were set,and the specific machining paths for the robot were obtained.Laser cladding experiments were conducted on the surface of the pit defects to verify the reliability of the 3D vision system proposed in this study.