A Maneuvering Target Tracking Method Based on RBPF

Abstract

The particle filtering (PF) is a widely used in present. However, as PF is used in nonlinear systems, a large number of particles are necessary to maintain high tracking accuracy and more computational burden are inevitably. Then, a new method of Rao-Blackwellised Particle (RBPF), in which Rao-Blackwell theorem is used to improve the performance of PF, is proposed in the paper. Firstly, nonlinear part and linear part of the maneuvering target tracking model are separated Based on Bayes principle. Then, the estimation of the linear part is dealed with by Kalman Filter (KF) and that of the nonlinear part by PF. The experiment results show that the tracking accuracy and the robustness of RBPF is higher than that of PF.