An adaptive trajectory prediction method for ping-pong robots

Abstract

Trajectory prediction is a key issue to a ping-pong robot, many algorithms have been developed. To get more robust and accuracy prediction under different serving conditions, this paper presents a new prediction method. The proposed method establishes two equivalent forms of the dynamic model of flying ball, where the discrete form for state estimation and the continuous form for trajectory prediction. The two forms share the same parameters' value. According to force analysis, It is found that the model parameters are deeply related to ball's state (position, velocity) So we train the model parameters offline respect to ball's state, instead of setting them to a constant value. This enables the model to be adapted accordingly online. Experimental results show the effectiveness and accuracy of the proposed method for the ball with different velocities. © Springer-Verlag Berlin Heidelberg 2012.