Adaptive UPF Based Interacting Multiple Model Algorithm for Integrated Navigation

Abstract

To resolve issues such as the degraded particle performance and difficulty in selecting the importance density function of particle filtering, this paper introduces an adaptive Unscented particle filter (UPF) based interacting multiple model algorithm. The method benefits from both the interacting multiple model filter and Unscented particle filter. It first uses the Unscented kalman filter to get the latest measurement information at k time for each model. And then, it takes interaction of the corresponding particle as input of each model, after matching the model and updating the probability of model and using residual re-sampling. Finally, it outputs interaction of the corresponding particle for each model. The proposed algorithm has been used in the GPS/DR integrated navigation system. Simulation results and their analysis demonstrate that the position and velocity error calculation of the proposed algorithm is much better than the Unscented particle filter, and the algorithm certainly improves the calculation accuracy of the GPS/DR integrated navigation system.