Design Of Ball And Plate System And Research Of Its Control System

Ball&plate control system based on computer vision is a nonlinear under-actuateddynamics system. It obtains the ball’s position through the camera and controls the motorwith the corresponding control algorithm. Its goal is to achieve the positioning control ofthe ball and trajectory tracking. Therefore it is a typical experiment platform of researchingintelligent control algorithms, which involves image processing, intelligent control, motioncontrol, machine vision and so on. So it has great significance to research it.The research object of this paper is ball&plate control system based on vision. First,the research situation of ball&plate in china and abroad, and composition of ball&platewere introduced. And based on the above analysis, the design scheme of the ball&platecontrol system bases on computer vision was realized. Second, according to the basicsituation of laboratory, the ball&plate control system based on computer vision wasdesigned, and the simulation platform was built. The property and functions of every partin the system were introduced. Third, the part of image processing about the ball&platesystem was designed. It included image preprocessing, such as gray-scale transformation,filter, and edge detection and so on, and the positioning of ball through Hough circledetection algorithm. The rectangle detection (tray detection) was achieved through contoursearching in preprocessed image and the position was approximating according to thegiven precision. On the basis of the previous experience and experiments of the controlsystem, the fuzzy controller was designed. The improvement measures for the main reasonwas put forward after observation of the control effect, and analysis of the reasonsreflecting control quality. The control interface of ball and plate control system in vs2008was designed. And based on the observation real-time data, error and the trend of motorcontrol value, it was found that in the construction of visual control system of ball&plate,due to assume that the board fluctuation is very small, there were some errors in detectionball’s relative position when collecting the ball’s position by camera. Fuzzy control was used in control the ball’s position and there is a serious lag in the whole system which leadto large fluctuation in the control process and large error of ball’s position. The ball justcan move in a certain range. The UKF (Unscented Kalman Filter) and AUKF (AdaptiveUnscented Kalman Filter) algorithm in the simulation environment to estimate the motionstate of the ball was designed to solve the problem of control delay due to system timedelay. The accuracy and precision of AUKF algorithm to predict the ball’s position wasverified through simulation which laid the foundation for the using of UKF and AUKFalgorithm in ball&plate control system.Finally, the positioning control was achieved in the designed ball&plate controlsystem based on computer vision. Though the expectation was not realized, the reason ofdifficulty in controlling the system was analyzed. The AUKF algorithm was proposed tosolve the main problems. And it needs further research to improve the precision of ball’sdetection. This paper accumulates some experience for design and control of ball and platesystem.