Research On Positioning And Grasping Technology Of Simple Shape Workpiece Based On Monocular Vision

To improve production efficiency and automation degree more and more machine vision technologies have been applied in some works such as targeted grasping,identifying,automatic sorting and so on.Aiming at the problem of simple-shaped workpiece's automatic loading and unloading,this paper proposes to obtain three-dimensional spatial information of simple-shaped workpiece which has the obvious geometric features by a monocular visual positioning system.Then the idea is proposed to guide the robot to grasp the simple-shaped workpiece by the information in this paper.This paper takes the common disk-shaped simple workpiece as the research object,and carries out the research on the workpiece positioning and grasping technology based on monocular vision.The working details are shown as followed:(1)In this paper,median filtering and grayscale transformation are used to preprocess the simple disk-shaped workpiece images,which effectively suppresses image noise and highlights the edges of the image contour.By comparing several common edge detection algorithms,the Prewitt operator is selected to extract edges from preprocessed workpiece images.A template library of simple disk-shaped workpiece is established and the shape-based template matching algorithms is used to identify the workpiece.To improve the matching speed and matching success rate of the matching algorithm,the matching threshold and pyramid layers are optimized.The recognition experiments demonstrate the method introduced in this paper can effectively identify disc-shaped workpieces from images.(2)Pinhole camera model is employed and the distortion correction step is introduced in camera calibration based on Halcon software to improve the accuracy in the paper.The three-dimensional spatial information of the disk-shaped workpiece is obtained by the plane positioning and depth information and the spatial posture information is obtained by the projection characteristics of the disk-shaped workpiece in the monocular vision system.The method is effective tested by the experiments andaslo can meet the needs of robot grasping.(3)In this paper,the "eye-to-hand" robot hand-eye calibration configuration model is employed,then a fast hand-eye calibration method is proposed to establish the conversion relationship between the camera coordinate system and the robot hand base coordinate system.An experimental platform for workpiece positioning and robot grabbing based on monocular vision was built,and the interface software between the vision positioning system and the robot motion control system was developed.Finally,it's successed to guide the robot to grasp the disc-shaped workpieces in different positions and different postures by the positioning information which is obtained by monocular vision.