Research On The Visual Guidance Method For Grasping The Special-Shaped Glass By Manipulator

In the industrial production line to improve the level of intelligence in manufacturing can be through machine vision related technology to complete the manipulator grasping the workpiece guidance system.Stability ?real-time and robustness and other performance is an important indicator to measure the visual guidance system.Hand-eye calibration and target identification positioning is the key point of the visual guidance system.Most of the time there is no calibration in industry site,using three coordinate meter is expensive.In addition,some industrial environment is so terrible that the target is not cognitive.The research object of this article is special-shaped glass and the center is not easy to obtain.There is partial absence of nonlinear deformation of image.In this paper,a selfcalibration method and a recognition and positioning algorithm are proposed for industrial requirements.The main contents are as follows:(1)A self-calibration method for non-fixed(eye-in-hand)calibration system is proposed.In this method,four linear independent translational motions,one actual rotational motion and one virtual rotational motion can be adopted,and the idea of extended focus is adopted without using the calibration reference.This method can by four linear independent translational motion,one actual rotation movement and one virtual rotation movement and based on extended focus without the use of calibration reference.Then,the camera internal parameter matrix,hand-eye relationship and target depth data of feature points can be solved linearly in turn.Experimental results show that this method is feasible and effective.By simply selecting feature points,the measurement errors of relevant position information data can be controlled within 0.40 mm under the premise that the robot body position remains unchanged.So it can meet the precision requirement of industrial field,and the application of hand-eye system in industrial production has reference significance.(2)An algorithm of contour recognition and location based on minimum external rectangle is proposed.The feature points of the contour segment are extracted and the weighted shape similarity calculated by rough matching and fine matching is obtained.And take the center coordinate of the rectangle as the center coordinate of the special-shaped glass(x,y).Theta is the difference in Angle between the edge of the rectangle and the origin coordinate of the image.Thus,the position information of the spcial-shaped glass can be obtained.Experimental results show that the proposed algorithm is real-time and stable.And it has some robustness against image loss nonlinear deformation robust to image loss and nonlinear deformation.The experiment proves that this paper presents a self-calibration method of non-fixed(eyein-hand)hand-eye system and a recognition and positioning algorithm based on the minimum external rectangle.Which can meet the requirements of general industrial production and application,and can be applied to the visual guidance system facing the manipulator to grasp the special-shaped glass.