Abstract: The inaccurate results of ball trajectory prediction lead to ping-pong robot can not play the table tennis intelligently.In order to reduce the trajectory prediction error,the following measures may taken:The proposed method first analyses the kinematics model of the flying rotary ball,and then constructs the motion equation and observation equation of the ball’s flying trajectory based on the Unscented Kalman Filter (UKF)．Finally ping-pong ball’s three-dimensional space position,velocity and angular velocity can be estimated,according to the three-dimensional space position information obtained by a visual observation system．The Matlab simulation experiments and real experimentals show that UKF algorithm compared to EKF algorithm in trajectory prediction time can be saved by 99%,and the tracking error is small.

Key words: Rotary table tennis,Trajectory prediction,Unscented kalman filter,Ping-pong robot