An improved resampling scheme for particle filtering in inertial navigation system

Abstract

The particle filter provides numerical approximation to the nonlinear filtering problem in inertial navigation system. In the heterogeneous environment, reliable state estimation is the critical issue. The state estimation will increase the positioning error in the overall system. To address such problem, the sequential implementation resampling (SIR) considers cause and environment for every specific resampling task decision in particle filtering. However, by only considering the cause and environment in a specific situation, SIR cannot generate reliable state estimation during their process. This paper proposes an improved resampling scheme to particle filtering for different sample impoverishment environment. Adaptations relating to noise measurement and number of particles need to be made to the resampling scheme to make the resampling more intelligent, reliable and robust. Simulation results show that proposed resampling scheme achieved improved performance in term of positioning error in inertial navigation system In conclusion, the proposed scheme of sequential implementation resampling proves to be valuable solution for different sample impoverishment environment.