机动目标跟踪的交互多模型泊松多伯努利混合滤波  

Interacting multiple model Poisson multi-Bernoulli mixture filter for maneuvering targets tracking

作　　者：陈壮壮 宋骊平[1] CHEN Zhuangzhuang;SONG Liping(School of Electronic Engineering,Xidian University,Xi’an 710071,China)

机构地区：[1]西安电子科技大学电子工程学院,陕西西安710071

出　　处：《系统工程与电子技术》2024年第3期786-794,共9页Systems Engineering and Electronics

基　　金：国家自然科学基金(61871301)资助课题。

摘　　要：满足共轭先验性质的泊松多伯努利混合(Poisson multi-Bernoulli mixture,PMBM)滤波器将目标状态分为泊松和多伯努利混合两部分,分别对这两部分进行预测和更新,具有较高的跟踪精度和较快的运行速度。在多目标机动场景下,使用单一模型不足以描述目标的运动,将导致跟踪性能的下降。针对这一问题,提出了一种交互多模型(interacting multiple model,IMM)PMBM滤波器,充分利用模型之间的交互信息,可以有效实现多机动目标的跟踪。同时,该算法采用序贯蒙特卡罗(sequential Monte Carlo,SMC)方法实现PMBM滤波,可应用于非线性场景。仿真结果表明,所提的IMM-SMC-PMBM算法可以有效地在非线性环境下跟踪数目变化的多机动目标,与IMM-SMC概率假设密度(probability hypothesis density,PHD)滤波器相比具有更好的跟踪精度和稳定性。Poisson multi-Bernoulli mixture(PMBM)filter,which satisfies the conjugate prior property,partitions the target state into Poisson and multi-Bernoulli mixture.The filter performs the prediction and update steps for those two parts separately and has fast operation speed while keeping high tracking accuracy.In the case of multi-target maneuvering,the single model is not enough to describe the target motion which will lead to a decline in tracking performance.To solve this problem,an interacting multiple model(IMM)PMBM filter is proposed,which can effectively track multiple maneuvering targets by making full use of interactive information between models.The proposed algorithm employs the sequential Monte Carlo(SMC)method to realize the PMBM filter,which can be applied in nonlinear scenes.The simulation results show that the proposed IMM-SMC-PMBM algorithm can effectively track multiple maneuvering targets with varying number in nonlinear environments.In comparison to the IMM-SMC-PHD filter,the proposed filter has high tracking accuracy and stability.

关 键 词：机动目标跟踪 交互多模型 序贯蒙特卡罗 泊松多伯努利混合 

分 类 号：TN953[电子电信—信号与信息处理]