用LWR学习进行乒乓球轨道预测的仿真研究

SIMULATION ON PREDICTION TRACE OF PING PANG BALL USING LWR LEARNING

摘要: 乒乓机器人的任务是将球击回到对方的场地内．这个任务由两部分组成：知觉和行动．本文研究机器人的知觉问题，也就是利用LWR（Localy Weighted Regression）学习的方法预测接触反弹后的球的轨道．

Abstract: The task of Ping Pong Robot is to hit back the ping pong ball to the opposite court., and it is made up of two steps: sensing and hitting. This paper studies the sensing of the robot. The method uses LWR(Locally Weighted Regression) learning to estimate the trace of the ping pong ball after bounding.