Research On Control Algorithm For Motion Tracking Platform

With the development of computer hardware and software technology, computer vision technology is getting more and more attention from people, as one of the important researches on computer vision, video moving target tracking has been widely applied in the fields such as military, aerospace, computer-aided design, intelligent robots etc. To simulate the gun control and achieve the locking and tracking on the target, PTZ is used to simulate the automatic gun turn able system. This paper researches in-depth on the extraction of the moving target area, target detection and tracking, Kalman prediction, and PID algorithm to achieve the control of PTZ.This paper firstly introduces the various control methods and their development of PTZ in stimulating gun control at home and abroad, then the major hardware and technical parameters. To achieve the target tracking, the Cam Shift algorithm is needed. In the tracking system, using the way of miss distance measuring to accurately determine the target location. Miss distance calculation units in the system will get through some calculating process such as the image acquisition, signal processing, tracking algorithms, storage, transmission etc. These operations will take at least 30ms of time, but the calculation cycle of the servo control system can make it to 0.01ms.This delay that is generated by miss distance calculations may result in large errors to the servo system. At the same time, if the control variables are only the miss distance of image processing it will cause the lack of the PTZ control bandwidth. Make great impact on the tracking, so some kind of algorithm to predict the miss distance is needed, using a variety of miss distance prediction through experiments. After comparing and analyzing, finally the research chooses the method of steady-state Kaman prediction, proving its characteristic of stability and real-time through the experiments. For there is no encoder feedback it can not correct the absolute position of PTZ. If the incremental PID algorithm is used there will continue to cause the accumulation of errors. PTZ will be out of control duding some time, so according to the technical conditions of this system, a separate PID control algorithm using integral is applied. Initially setting the parameters, PID tuning, researching on various tuning methods will make the PTZ adjusting time and overshoot level best. Finally the method of PID parameter turning based on curve fitting is choused to get the best effects. Experiments have proved the characteristics of effectiveness and real-time in the method that is proposed in this paper.