Archive | 2019
A New Multi-sensor Particle CPHD Filtering Algorithm for Bearings-only Multi-target Tracking
Abstract
Aiming at bearings-only multi-target tracking, a new multi-sensor particle CPHD filtering algorithm is proposed, which analyses the structure information of mixed linear/nonlinear state space models and combines particle filter and Kalman filter to predict and estimate the states of multiple targets to enhance the estimating performance of the PHD and cardinality distribution. The target state estimates are extracted by utilizing the kernel density estimation theory and mean-shift method. Simulation results are presented to demonstrate the improved performance of the proposed filtering algorithm. Introduction In bearings-only multi-target tracking, the number of targets is unknown and vary with time due to the uncertainty of target information. In addition, the problem of model nonlinearity caused by coordinate transformation of target motion modeling and measurement modeling, and the physical characteristics of passive sensors themselves, as well as the incompleteness of measurement information, all bring great difficulties to target tracking. How to track multi-target effectively based on bearings-only measurement information has always been a popular and difficult topic in both academic and engineering research[1]. Compared with other traditional multi-target tracking algorithms, probability hypothesis density (PHD) filtering algorithm based on random set theory can transform complex multi-target state space operations into single-target state space operations, effectively avoiding complex data association and combination problems in multi-target tracking[2-4]. The cardinalized probability hypothesis density (CPHD) filtering algorithm, which can make full use of the information of multi-target density and does not need to limit the number of targets to obey Poisson distribution, has attracted more attention. Many scholars have carried out relevant research[5-9]. In this essay, a new multi-sensor particle CPHD filtering algorithm is proposed which uses centralized fusion strategy. By mining the structural information of mixed linear/non-linear state model and combining particle filter (PF) [10] and Kalman filter (KF) to predict and estimate the state of each target, the PHD and cardinality distribution of multi-target can be better estimated. Simulation results show that the proposed filtering algorithm is effective in the challenging bearings-only multi-target tracking scenario. Problem Formulation Consider the following passive multi-sensor bearings-only multi-target tracking system: , 1 , , n k n k n k x Fx Gw \uf02b \uf03d \uf02b , \uf028 \uf029 , , , , , , 0 o n k k m k o m k m k h x v n z clutter \uf056 \uf056 \uf02b \uf03d \uf03d \uf03d \uf0ec \uf0ed \uf0ee , (1) Where , n k x is the system state vector of target n at k time , \uf028 \uf029 \uf028 \uf029 \uf028 \uf029 \uf028 \uf029 \uf05b \uf05d T , , , , n k n n n n x x k x k y k y k \uf03d \uf026 \uf026 , , n k w is a white Gaussian noise with zero mean and covariance matrix n Q , \uf07b \uf07d , , 1, , o m k k z m M \uf03d \uf04c is the set of target bearings generated by sensor o at time k which contains k C clutters, 2nd International Conference on Electrical and Electronic Engineering (EEE 2019) Copyright © 2019, the Authors. Published by Atlantis Press. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Advances in Engineering Research, volume 185