改进的UKF乒乓球轨迹预测算法

Improved UKF table tennis trajectory prediction algorithm

传统的无迹卡尔曼滤波(UKF)轨迹预测算法在乒乓球运动状态发生较大改变时,短时间内的预测结果会出现偏差。本文提出了一种基于UKF和简单运动学模型(SPM)的改进UKF轨迹预测算法,通过UKF对碰撞前轨迹进行预测,并得到乒乓球落点位置及运动状态量,进而通过碰撞反弹模型及运动学模型预测碰撞后的乒乓球轨迹。改进的算法在搭建的乒乓球机器人系统中进行轨迹预测,并与传统的UKF算法进行对比后可知,本文提出的改进算法较好地减小碰撞后的轨迹误差,算法简单,对硬件要求较低,可以实现实时预测,且具有较好的鲁棒性。

When large change exists in the trajectory during the motion of table tennis, the prediction results from UKF in a short time will be biased. In this paper, an improved UKF trajectory prediction algorithm based on UKF and simple sport model(SPM)is proposed. The UKF is used to predict the trajectory before the collision, and the position of the table tennis ball′s landing point and the motion state quantity are obtained, and the collision rebound model and the kinematics model are used to track table tennis. The improved algorithm is tested on the trajectory of high-speed moving table tennis in real scenarios, the results are compared with the prediction results of the traditional UKF algorithm. The experimental results show that the proposed improved algorithm can better reduce the trajectory error after collision. The overall algorithm is simple and has low hardware requirements. It can realize real-time prediction and has good robustness.