Study On Multi-target Recognition And Crawling Position Confirmation Of Handling Robot Based On Visual Guidance Technology

Currently, the robot has been widely used in various important aspects of the national economy. As computer vision technology and the rapid development of parallel image processing technology, the combination between image processing technology based on visual guidance and the robots are becoming closer. Visual guidance combined with technology and robots, the robot can be better used in industrial production, meeting the requirements of the diversity of the production of products. To solve the problem of robot crawling different workpoeces on the same production line, this paper uses the machine visual guidance technique to study a variety of different workpieces in target identification and robot crawling position depending on the workpiece conversion problems, and it is of great significance to the mixed production line.The specific work is as follows:(1)Designing the visual guidance system platform to build the mathematical model among the robot,the camera and the work in coordinate system:a matrix determined relationship between the camera and the workpiece coordinate system,taking advantage of Zhang Zhengyou calibration method;according to the spatial geometric knowledge and by making use of the robot TCP tool,we obtained the transformation matrix between the robot and the workpiece coordinate system.we used the workpiece coordinate system as the middle part of the conversion,and finally we determine the conversion between the robot and camera coordinate system,which provides mathematical support for robot’s crawling artifacts.(2) Studing the template matching method to identify multi-target.For template matching with a rotation characteristics of the target workpiece, using Hu moment and Zernike moments as a template matching method and compared the advantages and disadvantages between them, finally selecting the Zernike moments,with a superior performance and orthogonal invariance,as the target recognition approach.To improve the speed of target recognition,before making template matching,using the circular projection value to screen the reference points of the image matching.(3) Establishing a mathematical model of the robot D-H,according to the following parameters provided by the robot initial state to study the robot crawling position. Matlab software simulation of the robot DH model by entering joint angle values and the robot end effector in the initial state crawling position in space parameters are verified by the initial state of the robot in both forward and inverse kinematics solutions,which provides a theoretical support for the subsequent trajectory planning.(4) Upon completion of the above work,using the EPSON-C3 robot as the actuators and CCD camera as a visual guide to complete the work on a variety of crawling experiments,which showed that: This multi-target recognition technology has a positive effect to provide a new technology and new method to guide the flexible production of the robot.