Workpiece pose recognition method based on quadric surface features

Abstract

The invention discloses a workpiece pose recognition method based on quadric surface features, and the method comprises the steps: segmenting a quadric surface from a model, calculating a quadric surface feature parameter and a centroid distance, setting the type of the quadric surface as a vertex, setting a spatial geometric relation as the weight of an edge, and constructing a model subgraph; aquadric surface is segmented from the scene point cloud, calculating characteristic parameters of each quadric surface, setting the type of each quadric surface as a vertex, the vertex with the centroid distance meeting a set condition being connected with an edge, setting the confidence coefficient of the spatial geometrical relationship as the weight of the edge, and constructing a scene mothergraph; finding out quadric surfaces corresponding to each other in pairs through matching of the scene mother graph and the model sub-graph; performing spatial geometric transformation on the quadricsurfaces corresponding to each other in pairs to obtain a rigid transformation matrix of the model for each target object in the scene. According to the method, the poses of the multiple workpieces inthe scene can be rapidly, accurately and simultaneously recognized, and the method is suitable for the situation that different parts are stacked in the same scene.

Description

technical field [0001] The invention relates to the fields of intelligent manufacturing and robot control, in particular to a workpiece pose recognition method based on quadric surface features. Background technique [0002] In modern manufacturing, assembly is an indispensable link, and it is also the production process that takes the most time. Realizing robot intelligent assembly is the key to improving production efficiency. In the process of "pick-and-place" in the box, industrial robots encountered When it comes to the recognition of scattered and stacked workpieces, it is often powerless. Therefore, the pose recognition information is the key to guide the robot arm to pick up. Most of the existing pose estimation methods are based on local feature matching, such as SHOT, FPFH, PFH, 3DSC, etc. , they find the paired points with the same characteristics from the model and the scene, and calculate the rigid transformation matrix of the model for the target object in the ...

AI Technical Summary

Problems solved by technology

[0003] In order to overcome the defects and deficiencies existing in the prior art, the present invention provides a workpiece pose recognition method based on quadratic surface features, which solves the pose recognition problem of most industrial parts and can quickly and accurately recognize the The pose of multiple workpieces, and it is suitable for stacking different parts in the same scene, and the result is not affected by noise

Images