Improved square root cubature particle filter based navigation method for UUV

Nonlinear non-Gaussian system state estimation problem is widely exists in the field of UUV underwater navigation. In nonlinear non-Gaussian system, the analytical value of the posterior function is needed to approximate by the exact importance density function. The traditional particle filter (PF) adopts the state transition prior distribution function as an importance density function to approximate the posterior density function. For the lack of measurement information of the PF, it combines square root cubature Kalman filter (CKF) with strong tracking filter (STF), and proposes an improved square root cubature Kalman filter (ISRCKF). The ISRCKF can real-time correct the filter gain and ensure good estimation ability by introducing fading factor and weakening factor. Then, it proposes an improved square root cubature particle filter (ISRCPF) based on ISRCKF resampling. The ISRCPF algorithm that incorporates the latest observations into prior update phase develops the importance density function through the ISRCKF, so that it is more close to the posterior density. Simulation test based on UUV sea trail data is carried out to verify the ISRCPF algorithm. The results show that the filtering accuracy of the ISRCPF is higher than the SRCPF, CPF and PF. Considering two characters, both the time cost and the filtering accuracy, the ISRCPF is a good choice to improve the particle filter algorithm in underwater navigation.