Research On Scale-span Positioning And Precise Docking For Transfer Vehicle Based On Machine Vision

The optical module needs to be installed in the clean room of laser experiment from bottom to top after long-distance clean transfer,so efficient transfer equipment is particularly important.The equipment needs to carry the module to realize positioning and precise docking after long-distance transfer.At present,there are the following problems in the process: the automatic positioning technology cannot accurately and roughly locate due to the interference error;there is no general pose measurement method due to the narrow space during docking.Therefore,based on the needs of high-efficiency transport optical module,aiming at positioning and measuring the docking pose,this paper studies the method of cross-scale positioning and precise docking pose measurement based on machine vision,and verifies the feasibility of artificial rough positioning and pose measurement through experiments.The main contents of this paper are as follows:Firstly,a scale-span positioning method is proposed,which is based on coarse positioning and precise positioning.The function of the equipment is analyzed,and the overall scheme is designed,including the main structure and control system scheme,in which the visual positioning part is mainly studied.This paper analyzes the existing problems of rough positioning and docking pose measurement,puts forward the corresponding solutions,and studies the relevant theory,working principle and workflow of the method.Secondly,a rough positioning method based on panoramic image technology is realized.By generating 360 ° panoramic image,we can assist technicians to eliminate blind area of vision,observe and locate signs,adjust equipment in place,and solve the problem of inaccurate automatic coarse positioning.This paper compares several common image registration algorithms,analyzes the reasons why they are not suitable for this paper,and uses the registration method based on pixel mapping to achieve image registration.According to the color difference of different views in panoramic image,an improved image fusion algorithm with uniform color effect is proposed.Then,based on the coplanar circle feature,a non-common field of view measurement method is proposed.Based on the analysis of the current non cooperative target pose measurement methods and applicable scenarios,this paper introduces the method.The commonly used image preprocessing methods and image analysis techniques are studied,and the image processing steps which are beneficial to the pose calculation in this paper are obtained,and the ellipse equation parameters in the image are fitted.The formula of pose measurement based on space circle is deduced.By using the distance and angle constraints of features,the false solution of docking pose is eliminated and the correct pose is obtained.Finally,the experiments of rough positioning panoramic image and docking pose measurement are carried out to verify the feasibility of cross-scale positioning and precision docking pose measurement methods.The panoramic image generated by the experiment is clear and smooth,which can display the surrounding environment of the vehicle body and assist the driver to complete rough positioning.The laser tracker is used to measure the precise pose.The results show that when the distance between two coplanar holes is 1140 mm,the pose angle error is within 0.5 ° and the translation error is within 1mm.