Research on Optimized M-estimate Arithmetic in INS/USBL Integrated Navigation System

摘要

This paper presents a method in order to promote the robustness of INS(Inertial Navigation System) / USBL(Ultra-short Baseline Positioning System) Integrated Navigation System applied to the AUV under the influence of non- Gaussian observationnoise outliers, especially mixed Gaussian. The main purpose of the paper is divided in to two parts: 1) reconstructing the pseudo-observation based on coordinate transformation. Combined with the position error and attitude error information provided by SINS in real time, the observation equation is established. 2) in view of the outliers of observationnoise when the interference appeared through sound wave propagation, multipath effect and other conditions in poor underwater environment. Analyze the feasibility of eliminating outliers in M-estimation, an Optimized generalized maximum likelihood estimation (OM- estimation) arithmetic is proposed to solve the heavy-tailed (impulsive) observationnoise outliers. The simulation results show the OM-estimation filter arithmetic has a better performance to ameliorate the robustness for the INS/USBL with the observationnoise outliers obeyed non- Gaussian, especially mixed Gaussian.