Design Of Robot 3D-Vision Positioning System Based On Line Laser Profile Scanner

With the rapid development of China's economy and robot technology,robots are increasingly widely used in industrial production and daily life.In order to improve production efficiency and meet the increasing demand for automation in industrial production lines,the combination of machine vision technology and robot technology to provide visual positioning and processing guidance for robots has become a research hotspot in the field of robotics at present.In this paper,3D vision technology is applied to the field of robotics.A robot 3D vision positioning system based on line laser profile scanner is designed to study the positioning of complex surface objects in random poses in the working scene of the robot,and the experimental verification is carried out.The main research work is as follows:Firstly,the overall design scheme of the 3D vision positioning system of the robot in this paper is proposed,and the hardware platform of the system is built by using KUKA industrial robot and SmartRay line laser profile scanner in the way of "Eye-in-Hand" installation.Then,the method of using PC as the upper computer to control the robot motion and the method of using software and development kit in the process of system development are studied.In order to obtain the coordinate system transformation relationship between the line laser profile scanner and the robot end,a fixed point multi-pose method based on the standard ball was proposed for the hand-eye calibration of the system,and the hand-eye calibration software was developed with the GUI function of MATLAB.Then,XYZ-four-point method and ABC-world-coordinate method are used to calibrate the robot tool coordinate system,and the coordinate system transformation relationship between the robot end and the tool that installed at the end is obtained.The laser profile scanner driven by robot is used to scan the target object for many times,and the partial point cloud of the object is obtained from multiple perspectives.The partial point clouds are fused to complete 3D reconstruction to obtain the object's point cloud template,and the greedy projection triangulation algorithm is used to carry out surface reconstruction based on the point cloud template to obtain the object's 3D surface model.Then,the positioning principle of the robot 3D vision positioning system in this paper is described,and the object pose detection algorithm based on point cloud registration is proposed to obtain the coordinate system transformation relationship between the laser profile scanner and the target object in the working scene.The algorithm is verified through experiments,and the running results of the algorithm are visualized,and it can be seen that the point cloud template of the target object has a good registration effect with the point cloud of the current working scene,thus proving that the algorithm is feasible and effective.Finally,the positioning error in x,y and z direction and the straight-line distance positioning error of the 3D vision positioning system in this paper are obtained through the system precision measurement experiment,which proves that the system has a high positioning accuracy.Then,based on the object's surface model obtained by 3D reconstruction,the application of surface processing by robot is simulated through the object surface track point tracking experiment,and the positioning accuracy of the system is also verified indirectly,which proves that the system is feasible and effective.