Research On 3D Visual Orientation Guidance Of Robot Based On CAD Model

With the rapid development of industry 4.0 and the continuous progress of artificial intelligence,machine vision has gradually become an indispensable part of the industry and life,machine vision show in various forms in our life,face recognition,visual inspection,visual positioning and so on,can be said that the application is very wide.Visual counterpoint guidance technology has high research and application value in industry.2D planar wizard has been very mature in industry,and 3D guidance has become a popular field.Based on the principle of structured light,3D reconstruction of object surface is used to generate point cloud information.The camera calibration mode,the selection of 3D point cloud template,and the robustness and speed of point cloud matching algorithm are studied.The calibration of camera is mainly based on the comparison of various calibration methods,and a semi-automatic calibration method is designed.The experiment proves that it takes account of the advantages of both automatic and manual calibration.Select the point cloud template,in view of being guided in the production process there will be differences,and this difference is not a stable Gu is difficult to determine a good point cloud point cloud template,we create the entire product through the CAD model,as the intermediate template to it,to enhance the stability of the matching.Point cloud matching algorithm to improve the robustness and speed,the existing algorithm of point cloud matching speed and matching accuracy of defects,this paper proposed an image matching algorithm based on the gradient of NCC,through the experiment has obviously improved the speed,accuracy,also have certain improvement.