Research On Image Processing And Trajectory Planning Of Coupler Sand Cleaning Robot

With the continuous expansion of the scale of manufacturing industry,intelligent manufacturing technology and system has been widely and rapidly applied,especially the industrial robot as a typical automation equipment of "machine generation" has been applied to various fields.Because of the heavy labor intensity and the serious dust pollution,the manual operation can’t guarantee the health of workers and the production efficiency of sand cleaning.Therefore,it is of great significance to study the technology of robot system of casting sand cleaning based on vision.This paper focuses on the visual processing technology of casting cleaning and robot trajectory planning technology of casting cleaning robot system.Firstly,taking the coupler casting of railway freight car as the research object,the structural characteristics of coupler casting are analyzed,and the position,size and shape information of coupler casting are extracted by using visual image processing technology,which provides data for robot to carry out sand cleaning operation and track planning.In image processing,image enhancement,threshold denoising and feature information extraction are mainly used to extract the feature information of coupler casting surface.At the same time,aiming at the problem of weak image processing and insufficient continuity of traditional edge algorithm for coupler casting,an improved method of traditional edge algorithm is proposed to analyze the characteristics of coupler external contour,and the contour algorithm and the shape of coupler external contour formed by edge algorithm are compared and analyzed to improve the continuity of contour algorithm.On this basis,the optimization method of casting space geometry is put forward.Firstly,the feature points of the casting image are extracted and registered,and then optimized according to the geometric features to remove the interference information,so that the feature points are concentrated on the casting surface.According to the feature point matching,the disparity map is obtained,and then converted into three-dimensional coordinates,and the corresponding point cloud is formed.Based on the point cloud data,3D reconstruction of the coupler is carried out,that is,reconstruction of the coupler model with sand and part of the coupler model with sand.Part of the coupler with sand is scanned twice to get a complete coupler model,which can clearly reflect the sand cleaning process by reconstructing the three-dimensional model.The method of least square fitting contour and contour surface is used to restore the contour of the body near the sand cleaning feature of the casting.Then,according to the restored contour and contour surface,the trajectory planning of coupler riser,hole,surface and flash is carried out by using the subsection strategy,and the position and trajectory of the sand cleaning are marked.Finally,through the software simulation,it is verified whether the trajectory planning of the cleaning process of the coupler meets the requirements of the cleaning process of the coupler casting.Through research and analysis,it is considered that the robot system based on vision can complete the sand cleaning operation of coupler casting,and the image processing method and path planning method adopted are feasible.