在线角速度估计的乒乓球机器人视觉测量方法  

Online angular velocity estimated visual measurement for ping pong robot

作　　者：张远辉[1,2] 韦巍[2] 

机构地区：[1]中国计量学院机电工程学院,浙江杭州310018 [2]浙江大学电气工程学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2012年第7期1320-1326,共7页Journal of Zhejiang University：Engineering Science

基　　金：国家"863"高技术研究发展计划资助项目(2008AA042602);浙江省人才基金资助项目(R105341);教育部基本科研经费资助项目(1A4000-172210203)

摘　　要：针对目前乒乓球机器人在视觉跟踪过程中无法准确识别和跟踪旋转球轨迹,导致预测结果误差较大的问题,提出基于在线旋转角速度估计的视觉测量方法.该方法应用空气动力学的理论知识对旋转球的受力情况进行分析建模,构建旋转球轨迹的过程方程和观测方程,利用非线性扩展Kalman滤波器对包括角速度在内的运动状态进行估计.通过仿真实验和实际轨迹跟踪实验验证了该方法的有效性和正确性,且预测结果优于同类跟踪方法.该方法亦可应用于实时高速目标跟踪的场合.Currently, the failure of precisely identifying and tracking the spin ball trajectory in ping pong robot＇s vision system leads to large errors in the predicting result. Aiming at the problem, an online non- linear Kalman filter based visual measurement method was proposed to estimate the angular velocity. Aer- odynamics theory was applied to analyze the forces acting on the spin ball. The motion equation and obser- vation equation of the trajectory were constructed, and the nonlinear extended Kalman filter was integrated to estimate the motion states involving the angular velocity. Both simulation and actual experiments veri- fied the effectiveness and correctness of the method. The prediction result outperformed other tracking methods. The method can also be employed in the high-speed motion tracking conditions.

关 键 词：乒乓球机器人 KALMAN滤波 马格努斯力 角速度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]