Development And Design Of Robot Vision System For Weighing Instrument Verification

Weighing instrument verification has a quite complicated process.The verification process is usually manual operation.The weighing instrument needs to be verified at the factory and for a period of time after leaving the factory,which makes the calibration of the weighing instrument time-consuming and laborious.Therefore,it is of great practical significance to design a robot system that can automatically perform weighing instrument verification.This subject aims at the accuracy of weighing instrument verification requirements,the particularity of the process,and the intensity of manual labor in the verification process.At the same time,considering the intelligence of the verification process,a system of dual robots with visual collaborative operation is proposed.The design of the vision system mainly completes the high-precision threedimensional positioning of the weighing instrument,and then converts it into the basic coordinate system of the robot through various posture coordinate transformations and guides the mechanical arm to autonomously grasp,place and press the target object.The specific research content of the subject mainly focuses on the following aspects:(1)The subject researched the development status of verification automation devices at home and abroad.After analyzing the advantages and disadvantages of each verification system,the overall framework of the weighing instrument verification robot was proposed in this paper,and various types of weight gripping tools were determined.Lens and other related equipment selection.(2)For the two major types of standard weights used in weighing instrument verification,the subject proposes an image processing method suitable for the characteristics of the weight itself,which is used to extract the main target of the two-dimensional image.For F2-level cylindrical weights,due to lack of texture and large environmental impact,it is impossible to use SIFT and other operators for registration.It is proposed to use YOLOv3 algorithm for target type recognition,OTSU segmentation under backlight conditions and least squares fitting circle and other algorithms to separate subjects.For the lock weight,the edge detection method with high contrast background and the external minimum rectangle algorithm are used to determine the center.At the same time,the angle measurement of the grab bar of the lock weight should be determined by the angle between the center line of the external circumscribed rectangle long axis and the image coordinate axis.Finally,we compare other similar algorithms to prove that the algorithm used in the paper has certain advantages over other algorithms.(3)In order to obtain the depth measurement value of the target,this subject proposes a monocular camera to measure the depth value of the three-dimensional coordinates.The depth value of the center point of the target object is obtained by simulating the principle of binocular distance measurement with a monocular camera under the traction of a mechanical arm.The advantage of this algorithm over the previous binocular ranging algorithm is that the equipment is simple and there are no restrictions such as distance.At the same time,this method has the same accuracy as binocular camera ranging,but it has obvious advantages in complex systems with stringent cost and space requirements.(4)For the key part of the verification process,this topic proposes to use a small six-axis robot for this operation.A small vision system is built on the robot to identify the key position.Because the keys at the table are all OCR,the system mainly uses OCR recognition of region template matching,used to locate the two-dimensional image location.Threedimensional depth coordinates are still measured using a monocular camera.(5)By performing camera internal parameter calibration and camera calibration in the robot base coordinate system,the three-dimensional coordinates in the image coordinate system are converted to the robot base coordinate system.And verified its accuracy and feasibility in the experiment.(6)After completing the relevant main content research,this subject uses the VS2017 compilation platform to design the visual host software.The main functions of this software are image acquisition and real-time display,image processing result display,target data display at each coordinate,and robot arm movement,RPY,three motion modes of joint),real-time display of the coordinate value of the end of the robot arm,and the opening and closing of the air pump and the grip motor.Finally,the visual system was installed in the SIASUN robot for experimental verification.The experimental results show that the positioning error of the two-dimensional image is less than 1.5mm;the error in the measurement of the depth value is less than 10 mm,and the detection target accuracy rate is about 92%.Accurate crawling and key operation proved the validity and reliability of the research content in the subject.