Trajectory Tracking Of Plate Cutting Based On Binocular Vision

With the development of the industrial process, the desire for automation is becomingmore and more urgent. Accurate and efficient robots are widely used in industrial production.Cutting robot is widely used in nuclear power, shipbuilding, chemical, automotive and otherindustries. At present, the mostly used methods in cutting robot’s programming are teachingplayback and off-line programming. The problem with these methods is that we needreprogramming or teaching over again when the workpiece to be processed changes itsposition or some other variable changes. Furthermore, we can’t detect the real-time quality ofthe cutting process. In this paper we made good use of the advantages of off-lineprogramming and at the same time visual feedback is introduced to transport the real-timecutting quality. With all these done, we have gotten quite satisfactory cutting quality. With thebackground of practical projects, against the trajectory tracking problem coming from thecutting process for large plate, the author does some studies on the relative trajectory processand trajectory information acquisition.The hardware selection and build of Binocular Vision Tracking System is done. Detailsof each element and its role in the system are given. With all these done, the system can beused for the future studies and experiments.First the Binocular Tracking System is chosen and the basic principles of stereo vision areexplained. Then the author gives details of the camera imaging principle and the method ofimage matching and image correction. The method of getting three-dimensional informationof the spatial point is showed on the base of the calibrated image. By contrast, CalibrationPrinciple of Zhang is selected in consideration of the actual requirement of the project. Basedon the method, the author does some corresponded studies and improvements.For the design of steel cutting trajectory algorithm of image processing, after noiseprocessing, image enhancement is used to improve rail line contrast. Against the differencesin light, adaptive Otsu segmentation algorithm is proposed, and the author gives themorphological processing approach.After image preprocessing, several trajectory extraction algorithms are given. This articleuses “hough” transformation to determine the initial position, and obtains the precisetrajectory information using the least-square method. Different trajectories were extracted forexperiments.And also, the trajectory extraction algorithm is verified. What’s more,weanalyzed the special interference circumstances encountered in the steel trajectory extraction,such as arc, rust and iron of the edge. In the end we extracted the trajectory. Finally, the Straight Mouth, Uphill Mouth in the actual cutting are tracked. Afterextracting the check line and the cutting edge, the paper gives their three-dimensionalreconstruction,with renderingsand trajectory parameters. Trajectory deviations in some casesare summarized too. The given traction algorithm is passed to the MCU through serialcommunication to adjust the offset.Experiments showed that the theory of the proposed binocular cutting trajectory trackingsystem, the Image processing algorithm and the tracking strategy are practical. It can track thecutting edge accurately.By adjusting the cut bias, it can also improve the control accuracy ofthe cutting robot.