Study On Workpiece Grabbing System Based On Binocular Vision

With the modernization and intelligence of industrial manufacturing,the combination of machine vision technology and robots can improve the ability of industrial robots to perceive and adapt to complex environments,and realize the intelligence of robot grabbing operations.Aiming at the sorting problem of stacked workpieces,the workpiece grabbing system based on binocular vision is studied,which is mainly divided into four parts:(1)Scheme design of work piece grasping system.The hardware components of the workpiece grab system are compared and demonstrated,and the functional modules of the system are designed.The principles of single-target calibration,dual-target calibration and hand-eye calibration are introduced to provide a theoretical basis for calibration experiments of visual systems.Aiming at the problems of low contrast and random noise in images,the common image preprocessing algorithms are compared and analyzed.(2)Research on workpiece recognition algorithm.The workpiece recognition algorithm with similar shape and complex texture is studied.To solve the problem of long recognition time of SURF algorithm,FREAK operator is used instead of SURP operator to describe and match feature points,which shortens the matching time of feature point pairs.After the coarse matching of the feature point pairs,the RANSAC algorithm is used to fine match them,which improves the matching accuracy of the feature point pairs.The comparison experiments of the three recognition algorithms show that the algorithm in this paper has better accuracy and real-time performance,and has good robustness.(3)Research on workpiece positioning algorithm.According to the real-time and accuracy requirements of the application scene,the SGBM algorithm is used for stereo matching to obtain the depth information of the target workpiece.A method based on three-dimensional model is designed to estimate the attitude of the target workpiece,and a workpiece pyramid template library is designed to shorten the positioning time of the workpiece.Take the matching degree and position height of the workpiece as the evaluation index of the grab priority.Through the positioning experiment of the workpiece,the feasibility of the positioning algorithm in this paper is verified.(4)Implementation and verification of workpiece grab system based on binocular vision.The hardware selection and program design of the system are introduced.The calibration parameters of the binocular camera are solved through calibration experiments.A gripping experiment for stacked workpieces is designed to verify the feasibility of the workpiece gripping system,which meets the actual use requirements.The experimental results show that the maximum position error of the target workpiece in the X-axis and Y-axis directions is within 1mm,the maximum position error in the Z-axis direction is within 2mm,and the maximum error of the attitude angle is within 3°,which can meet the actual gripping requirements of the system.