Method and system for identifying and positioning targets on basis of three-dimensional point cloud

Abstract

The invention discloses a method and a system for identifying and positioning targets on the basis of three-dimensional point cloud. The method and the system have the advantages that procedures for identifying and positioning target workpieces during industrial production feeding and discharging are divided into an offline feature extracting module and an online feature matching module, accordingly, the computational complexity in real-time phases can be reduced to a great extent, and the arithmetic speeds can be increased; matching quality estimation is introduced to obtain accurate three-dimensional locations and gestures of the workpieces, and accordingly industrial robots can grab the identified workpieces in corresponding gestures under the control; the method and the system are different from target identifying and positioning algorithms customized and developed for local geometric shape features (such as circles and holes) of existing workpieces, the targets can be universally identified and accurately positioned by the aid of the method and the system, accordingly, algorithms do not need to be redeveloped for the different workpieces, and the development time can be saved; the problem that the existing workpieces are required by the traditional feeding and discharging schemes for robots to be fed into visual fields of existing cameras according to certain gestures can be solved, the method and the system are free of optional requirements on the gestures of the workpieces in visual fields of cameras, and the workpieces can be arranged in a scattered manner or can be stacked on one another or can be shielded by one another.

Description

[0001] Technical field: [0002] The invention belongs to the technical field of machine vision, and in particular relates to a target recognition and positioning method and system based on a three-dimensional point cloud. [0003] Background technique: [0004] In the industrial production process, loading and unloading is an indispensable and frequently occurring production process. In the current industrial production process, different loading and unloading methods are adopted according to the volume and weight of the incoming materials, such as using a vibrating plate to arrange the workpieces (such as screws) with small volume and weight in an orderly manner and supply them to equipment such as locking screw machines; For workpieces that are medium in size and weight and easily damaged, they are manually picked and placed; for workpieces that are large in size and weight and in a harsh production environment, machine vision is used to guide industrial robots to load and u...

AI Technical Summary

Problems solved by technology

Whenever a production workpiece is replaced, it is necessary to develop a corresponding recognition and positioning algorithm based on the local geometric characteristics of the workpiece itself. The current vision guidance scheme for industrial robots lacks versatility for different workpieces, which seriously limits production efficiency.

In addition, in the currently implemented machine vision-guided industrial robot loading and unloading scheme, the workpiece needs to enter the camera field of view according to a certain posture, and only the position deviation of the workpiece on the XOY plane and the deflection angle around the Z-axis relative to the standard posture can be obtained, which cannot be completely Obtain the 3D position and attitude information of the workpiece

Images