Research On Image Recognition And Motion Control Of Dispensing Robot

In the manufacturing industry of semiconductor processing machine in today's China,most of the dispensing operations of work-pieces are still done manually,which is inefficient and unfavorable to human health;and a part of the market share also is occupied by manual teaching industrial robots,but when there are many types of work-pieces or when the work-piece is offset,the teaching robot is much less efficient.The market prospect of dispensing robots with machine vision capabilities that automatically identify and position work-piece is particularly impressive.This paper starts from the actual cooperation of the factory,builds a dispensing robot system with visual function,designs the corresponding image processing algorithm,successfully extracts the trajectory information,and completes the coordinate of the vision system and the robot through calibration,thus realizing the transformation of the trajectory information.And the transmission,and developed the system software platform,and finally completed the simulation and system debugging of the robot.The main research contents of this paper are summarized as follows:1.Program development and construction of the visual platform.Starting from the needs of enterprises,understanding the development status of related fields at home and abroad,designing a dispensing robot system that meets the actual situation,which can identify and locate different work-piece,and formulate a solution for dispensing multiple planes.Considering the research object and the corresponding experimental environment,the selection of the device and the construction of the visual platform were completed.2.Camera parameter calibration.In order to establish the connection between camera pixel coordinate system and world coordinate system,the imaging mechanism of industrial camera is deeply studied,and the conversion relationship between four coordinate systems is derived.The method of nonlinear model parameter solving is proposed.Zhang Zhengyou's calibration method and its solution ideas were studied.The camera calibration experiment was completed and the internal and externalparameters and distortion coefficients of the selected camera were obtained.3.Image processing and feature extraction.A variety of algorithm programs such as low-pass filtering,Gauss filtering and median filtering are programmed to process the acquired original image,and the best filtering method is selected.Comparing the results of various combinations of expansion and corrosion,the edge detection processing of the image is performed,the image contour is extracted by multiple iterations,the arc and line features of the image are detected by Hough transform,and the coordinates of the trajectory are successfully extracted,and the characteristics of the template are adopted.The matching obtains the basic parameters of the glue tank.4.FANUC robot programming and control.Master the FANUC robot programming language and control method,complete the zero calibration of the robot system and the hand-eye calibration between the camera,and realize the transformation of coordinates.The kinematics analysis of the six-axis robot of this model is carried out,and the physical model is established.5.Complete the robot motion simulation.Robots and related 3D object models were built in Roboguide simulation software to restore their operating environment.A method of automatically generating a control program by directly importing the trajectory is developed,which greatly simplifies the operation process and successfully completes the motion simulation and program conversion.6.Design system software and perform system debugging.The C# language is used to design a simple and practical software for the system,which can realize data conversion,process control,data storage and status display.And complete the system overall debugging,the robot can locate the track and move according to the dispensing track,the system can be applied to the actual dispensing operation.