3D Pose Estimation Method in Unordered Grasping Based on Shot and ICP Algorithms

Abstract

The present invention relates to a three-dimensional attitude estimation method in disorderly grasping based on SHOT and ICP algorithms. The method is used for object pose estimation in the disorderly grasping process, and the SHOT rough matching method provides a good initial stage for ICP fine registration. Position, the construction of the M matrix in the SHOT algorithm, the eigenvalue decomposition, and then construct the local reference system; calculate the topological features under the corresponding reference system, and save the results in the histogram, and then extract the corresponding feature results of the model and scene point cloud Carry out Hough transform, eliminate wrong point pairs, and calculate to obtain the final rough registration pose estimation result; find the nearest point of ICP algorithm, construct the covariance matrix, and solve the rotation matrix and translation vector that minimize the objective function, solve The traditional ICP is applied to the problem that the registration accuracy drops when the point cloud with holes is collected due to the reflection of the workpiece.

Description

technical field [0001] The invention relates to the technical field of three-dimensional attitude estimation, and mainly relates to a three-dimensional attitude estimation method in disorderly grasping based on SHOT and ICP algorithms. Background technique [0002] At present, there have been many research results on image-based object recognition and attitude estimation, but the imaging process of two-dimensional images is mapped from three-dimensional space to two-dimensional space, which will lose a lot of information; and a fact that cannot be ignored is that the most A good vision system should be oriented to the three-dimensional world. With the improvement of automation, the requirements for computer vision systems in many fields such as robot navigation, industrial parts detection and grasping are getting higher and higher. Therefore, object recognition and pose estimation based on two-dimensional images can no longer meet the requirements of human beings. At presen...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of object pose estimation in the disorderly grasping process, the present invention provides a three-dimensional pose estimation in disorderly grasping based on SHOT and ICP algorithms with strong anti-noise ability, high robustness and high precision Method for 3D object recognition and pose estimation in industrial sites

Images