A method for locating engine valve grabbing points

Abstract

A method for locating engine valve grabbing points, which acquires 3D point clouds and corresponding 2D images; obtains overall 3D point clouds through 3D point clouds and overall oblique frame filtering with oblique frames, and performs regional growth clustering on the overall 3D point clouds , segment the car engine valves by adjusting the smoothness parameters of the region growth clustering; perform image PCA processing on the 3D point cloud of the rod to obtain the position and attitude of the grabbing point; determine the corresponding The top slanted frame, the top 3D point cloud is obtained by filtering the slanted frame, and the center of gravity of the top 3D point cloud is calculated to obtain the top center of gravity of the car engine valve; the top center of gravity is obtained by projecting the obtained top center of gravity point onto the x vector of the initial grab point attitude The projection point, determine the direction of the x vector of the initial grab point attitude; calculate the y vector of the initial grab point attitude according to the x vector and the z axis of the initial grab point attitude; use the x vector and y vector to determine the initial The z vector of the grab point pose.

Description

technical field [0001] The invention relates to the technical field of grab point positioning, in particular to a method for locating an engine valve grab point. Background technique [0002] With the improvement of industrial automation technology, the development of automobile automation production is getting faster and faster, and the automatic grabbing of automobile engine valve parts is also required. In order to facilitate grabbing, the grabbing point of the engine valve is generally required to be on the valve stem, and the grabbing posture should be consistent with respect to the top of the valve. [0003] In 3D vision, due to the reflective characteristics of the engine valve itself, the point cloud collected by the camera hardware is often incomplete, especially the top of the valve is not only affected by reflection, but also affected by the imaging characteristics of the 3D camera point cloud. Cloud quality is often severely lacking, which makes it extremely dif...

AI Technical Summary

Problems solved by technology

[0003] In 3D vision, due to the reflective characteristics of the engine valve itself, the point cloud collected by the camera hardware is often incomplete, especially the top of the valve is not only affected by reflection, but also affected by the imaging characteristics of the 3D camera point cloud. Cloud quality is often severely lacking, which makes it extremely difficult to locate grab points

[0004] The traditional engine valve grab point location often adopts the point cloud European clustering method to segment the valve point cloud, and then performs principal component analysis (Principal Components Analysis, PCA) and other processing on the segmented point cloud to obtain the grab point pose. This method can achieve better results when the parts are placed relatively scattered, but when the parts are placed relatively compactly and the point clouds between the parts are connected together, the point cloud is difficult to separate, and when the point cloud at the top of the valve is incomplete , the calculated position and attitude of the grasping point will have a certain deviation, resulting in incorrect positioning of the grasping point

Images