A novel trajectory prediction approach for table-tennis robot based on nonlinear output feedback observer

It is a key issue for table tennis robot that the flight trajectory of ball could be rapidly predicted within a certain precision. Due to the nonlinear model of the ball´s flight, only all state including the position and velocity at each event are catched, could its whole trajectory be obtained. However, since only the position could be measurable by cameras, the velocity should be predicted by the measured position. To solve this problem, this paper employs a novel output feedback observer to the ball´s flight trajectory prediction for the velocity estimated. The stability and performance of the observer with exponential convergence is proven by Lyaponov method, and conditions for the stability of the observer are given. Compared with the traditional method, the approach proposed in the paper has some advantages of fast convergence and high precision. Finally, he experimental results shows the performance of the proposed approach.