扩展概率数据关联新算法

基于新的联合事件定义，推导出一种联合概率数据关联新算法．在此基础上，分析了该算法（本文取名为ＥＰＤＡ）与概率数据关联ＰＤＡ的关系，从而得出ＥＰＤＡ可看成ＰＤＡ的一种扩展的结论，同时，还从理论角度说明Ｆｉｔｚｇｅｒａｌｄ提出的简化ＪＰＤＡ算法中近似经验常数可精确计算，也可近似给定，不仅仅只能靠经验给定．ＥＰＤＡ克服了传统ＯＤＡ算法不可回避的组合爆炸难题，是一种性能优良的数据关联算法．仿真分析证明，本文提出的扩展概率数据关联算法实时性、有效性强，适于工程应用．

A New Probabilistic Data Association Algorithm

We,offering a new joint event definition after careful consideration, present a new jointprobabilistic data association called extended PDA (EPDA). EPDA compares favorably with Cheap - JPDA, which overcame the combinatorial explosion unavoidable in using JPDA.Cheap - JPDA proposed by FitZgerald 3] appears to suffer from two defects: B in hismarginal probabilistic formula is crudely treated as a constant and many relevant fsctors aredisreSarded' We prove that B is not a constant I it depends on many factors, and we presenteq. (13) for its eomputation. The key equation for EPDA are eqs. (12) through (14). Data for an illustrative example are given in Table 1. Table 2 gives the simulation results of tallexample when EPDA is used. In Table 3, Tesults obtained with EPDA are compared with those obtained with PDA, Cheap- JPDA, and JPDA. In Table 3, the second row gives the time required for one Monte Carlo simulation, and the third row gives the number of times that tracking is unsuccessful. Table 3 clearly shows our new EPDA is better than existing data association algorithm.