Research On Workpiece Positioning And Robot Grasping Technology Based On Monocular Vision

It has important application value to improve the degree of automation and production efficiency using robot and machine vision technology in a modern factory,especially when robot technology and machine vision technology were combined into a system which has autonomous decision-making capability of detection,processing or sorting.Based on the development of visual technology,this paper starts with the analysis of the characteristics of visual system and the advantages and disadvantages of monocular vision and binocular visual system were compared.Because monocular vision has the advantages of Simple structure and high performance-price ratio,this paper focuses on the research of the workpiece location and robot grasping technology based on monocular vision.The main ideas and innovations are as follows:Aiming at the problem that it is difficult to recognize 3D pose of workpiece using monocular vision,the research route of workpiece location and robot grasping technology based on monocular vision is established in this paper,which combines experiment and theory analysis.The experimental platform is built at the same time.In order to solve the difficult identification with large localization error problem which caused by the surface reflective of metal workpiece and texture feature lose,the image pre-processing and recognition schemes for the planar and 3D workpiece of metal materials are determined respectively through experiments in this paper.Aiming at the typical applications of monocular vision positioning and robot grasping guidance of the general workpiece on production line,three steps are studied: This paper studies monocular vision localization and robot grasping in three steps aiming at typical applications: firstly,for the application of positioning and grab of conventional plane workpiece this paper analysis the theory of monocular vision positioning on the plane workpiece start from the model of monocular camera imaging to determine the basic problem of monocular vision.Secondly,for general three-dimensional positioning applications,this paper put the positioning of nut which have a three-dimensional shape as an example,the feasibility of scheme that positioning of 3D workpiece using monocular based on the a priori constraints were analyzed.The proposed scheme realizes the monocular vision localization and robot grasping by recognizing the points in the calibration plane and obtaining its spatial position and combining the known prior constraints.The scheme has the advantages of Simple hardware structure and short software development cycle.Finally the shortcomings of above visual positioning and grasping system were analyzed,the scheme of workpiece positioning and grasping based on the CAD model was put forward.The positioning scheme includes two steps: the first step is to estimate the position and attitude parameters of the workpiece by template matching.Then the position and attitude parameters are taken as the initial parameters of the iteration,and the real position and attitude parameters of the workpiece are obtained by real-time iteration.The experimental results demonstrate the feasibility of the scheme,compared with other schemes of workpiece localization the scheme based on CAD model is more flexible,and this scheme provides a new idea for the research of monocular vision guided robot grasping system.