Robot 3D Vision Measurement Algorithm For Complex Surfaces

The production and manufacture of aeroengine is an important part of the aviation industry.The quality of aeroengine production directly determines the performance of the space shuttle.Therefore,the precise production and manufacture of aeroengine has always been a national research focus.Aeroengine Blades account for about a quarter of the engine share,so the quality of blade production has an important impact on engine performance.In this paper,the precision measurement and machining of Aeroengine Blades and other complex special-shaped surfaces are studied,mainly to solve the problems of low measurement accuracy,poor real-time performance and robustness in the measurement and machining process of Aeroengine Blades.This paper focuses on the point cloud registration algorithm of aeroengine blade in the process of 3D vision measurement,and applies it to the high-end intelligent manufacturing line,and achieves good registration measurement results.The contents and achievements of this paper are mainly composed of the following parts:First of all,the hardware composition,installation mode,software module and measurement process of the robot 3D vision precise measurement system in high-end intelligent manufacturing line are introduced.Among them,the hardware system mainly consists of industrial robot,industrial camera,three-dimensional vision sensor,light source and corresponding industrial lens.The software module is mainly divided into software layer,function layer and application layer;the precision machining and measurement part of blade is divided into the process of obtaining,preprocessing,registration and measurement of blade point cloud.Then,the point cloud preprocessing of the collected aeroengine blade point cloud is studied.The project of preprocessing mainly includes two parts: the simplification and filtering of blade point cloud.The aeroengine blade point cloud obtained by Conner vision 3D vision sensor has large environmental noise,and there are some outliers and burr points.After the experimental comparison and research of the filtering algorithm,this paper uses statistical filtering and bilateral filtering algorithm to preprocess the point cloud,filter the noise of the blade point cloud,and reduce the accuracy of the subsequent blade point cloud registration Interference.Blade point cloud preprocessing is to provide a good data set without redundant noise for blade point cloud registration.Then,this paper studies the point cloud registration algorithm for complex irregular surfaces such as Aeroengine Blades.After the research of existing point cloud registration algorithm for complex irregular surfaces such as blades,it is found that the current point cloud registration algorithm can not meet the requirements for blade point cloud in this experimental platform In order to meet the requirements of accuracy and rapidity of measurement,this paper first proposes a new algorithm based on multi-scale descriptors.According to the spatial structure characteristics of point clouds on complex and special-shaped surfaces,multi-scale descriptors are constructed to obtain the corresponding feature descriptors sub point pairs that can describe the features of point clouds on blades.Furthermore,an algorithm based on spatial angle filtering and false registration is proposed to remove the false registration pairs,so as to obtain the corresponding descriptors sub point pairs that can accurately describe the features of point clouds and surfaces,which greatly improves the performance of point clouds on blades Registration efficiency and accuracy.Experiments show that the algorithm can meet the requirements of blade measurement and machining accuracy in high-end intelligent manufacturing line,and has strong robustness.It has good practicability in the precise three-dimensional measurement of point cloud of complex irregular surface such as aeroengine blade.