Research And Application Of Algorithm For Badminton Path Tracking Based On Computer Vision

Badminton is an athletic sport which need of high real-time performance.To make the robot play with people,the controller should trace the speeding shuttlecock in real time.Such issues exist in the area of industrial and military widely.The main researches are presented as follow:(1)Comparison of three dynamic target tracking algorithms: background subtraction,optical flow and frame-difference method.With some experiments and the corresponding analysis,we elaborate their merits and demerits.(2)Build the aerodynamic model of badminton and analyze the effects of aerodynamics and gravity to the trajectory.(3)In the process of prediction of badminton trajectory,there exists system noise interference.To solve this problem,we calculate the maximum likelihood estimate of badminton states by Bayesian Estimation.We analyse and compare two nonlinear filters(UKF and PF)and confirm conclusion through the experimental result.(4)Build the background robot system which consist of visual part and executes part.The visual part is used to predict the where a shuttlecock will land and calibrate the attitude of the background robot.The executes part is used to executing commands which the visual part send.