Trajectory identification of spinning ball using improved extreme learning machine in table tennis robot system

The trajectory prediction and classification play an important role in table tennis robot motion control, and the methods and results of classification affect the success rate of robot´s strike. The success of strike to spin relies on the accurate identification of ball´s trajectory and the computation speed of the robot. This paper analyzes the problem of spin classification firstly. Then an adopted and improved the Extreme Learning Machine (ELM) model is presented. A rigorous theoretical proof of the improved ELM algorithm is also presented. In the condition of non-massive data, we validated that the improved ELM model can achieve a better precision of identification on spin.