The Work-piece Posture Recognition And Grabbing System Based On Monocular Vision

As the development of industrial robot technology,robots can be used in more and more producing and processing areas to take place of the traditional manual operation.It has become an important problem that how the robot can recognize the positions of targets with different shapes and sizes.Nowadays,we usually identify the location of targets with the method of binocular vision which needs high requirement of camera precision and high cost of hardware.This method is too complex to satisfy the requirement of industrial production.The design purpose of this article is to propose a method to recognize and crawl workpiece with industrial camera and infrared sensor to solve the problem of recognizing target.The hardware was selected according to the system design requirements to build up experimental platform composed by the visual sensor systems,motion control systems and industrial robots after researching the domestic and foreign machine vision technology and design technology roadmap.Then a equipment was designed to install the industrial camera and infrared sensor to adjust the distance between them precisely.After doing the camera and robot hand-eye calibration in eye-to-hand platform,the coordinates of the workpiece center was calculated from the outer contour of the image with the algorithm of dual-threshold Canny edge detection in the environment of VS2008 and OpenCV.After matching the workpiece and template image feature,the shape of the workpiece is determined by the SURF algorithm.The program was designed according to the characteristics of robot to make it recognize and grab three kinds of workpiece in different poses.The feasibility of the experimental program was verified by analysis and comparison of the experimental results.This article designs a work-piece recognition and grabbing system based on monocular vision and infrared sensing technology.This method costs less and the development cycle of software is short.It provides a new way for industrial robots to crawl.