Industrial Robot Path Recognition Based On Laser Scanning And 3-D Reconstruction

According to the complexity requirements of fast changing and growing of modern industrial,the robot is required to “inexhaustibility " to carry out simple and complex work.In the welding industry,it is very important to combine the vision sensor with the robot to form an intelligent robot,which is very important for the robot to track the weld seam accurately.In order to transform the feature points from the image to the 3D coordinates of the image,and adjust the deviation and get the information of the width and depth of weld pool,it is necessary to calibrate the vision sensor;In order to obtain the three-dimensional shape of the weld,it is convenient to analyze the welding quality,and it is necessary to reconstruct the welding seam.So the main research content of this paper is as follows:Light plane calibration.A new method of optical plane calibration is presented---semi-active calibration.Firstly,the internal parameters of the camera and the angle between the camera’s optical axis and the vertical direction are obtained by using the Zhang Zhengyou’s method.By using the mechanical mechanism controlling the camera and the laser at the same time,there is a plane parallel to the plane of the camera and the plane.Secondly,through the use of mechanical mechanisms to control the visual sensor(mainly camera and laser)to move two times in the vertical direction(two times up or down),it is proved that there is a plane parallel with the laser plane in the camera and the plane,which can be obtained from the method of the invisible plane,that is,the light plane method.The plane equation of the laser plane in the camera coordinate system is obtained by using the contact plane and normal vector.Also introduced an existing light plane calibration algorithm,and mutual position determined by a freely moving target plane collinear three feature points.Based on the model of perspective three points,according to the three feature points in the camera image plane pixel information obtained on the light plane calibration points in the camera coordinate system of coordinates.After multiple moving targets,the coordinates of multiple target points are obtained in the camera coordinate system,and the plane equation of the optical plane is obtained by fitting these points.And the two kinds of calibration algorithms are compared.The V butt groove weld is seen as the main object of study,weld tracking and measurement information by using the planar optical semi active calibration method,it is proved that the accuracy of the calibration method and meet the requirements of precision industrial.3D reconstruction.A new data compression algorithm is proposed---double deviation method.The principle is: for the three points: the left point,the middle point,the right point.The first is to calculate the Euclidean distance between the straight line of the middle point to the adjacent left and right sides.The second is the Euclidean distance between the straight line which is the right point and the center point.While the data points to meet the above two conditions are kept down.Firstly,Through the image acquisition to the minimum region of interest(ROI)determination,adaptive image segmentation,center line stripe extraction,filtering and other processing,the point cloud data.Secondly,the point cloud data obtained by the image processing is reduced by using the double deviation method,then the correct topological connection relationship between adjacent points is obtained by using the 2D parallel contour line processing.Finally,the MATLAB simulation tool is used to reconstruct the welding seam.the weld 3D model through double deviation point cloud data coding algorithm and 2D contour parallel line algorithm has important scientific and use value.