Underwater vehicle near real time state estimation

Abstract

Acoustic time-of-flight positioning schemes are widely implemented for aiding underwater inertial navigation systems. The ping-response protocol and asynchronous nature of the returns of long-baseline (LBL) systems do not satisfy the standard assumptions necessary for Extended Kalman Filter (EKF) solutions. This paper presents a Near-Real-Time (NRT) framework for LBL aided inertial navigation. The solution proposed herein implements an optimal Bayesian state estimator over the time-frame of each LBL transponding cycle. This Maximum-A-Posteriori (MAP) solution considers all navigation sensor information collected during each LBL cycle and is computed at the conclusion of the LBL cycle. The solution between LBL cycles is computed by standard extended Kalman filter (EKF) methods for all other measurements (e.g., Doppler velocity log (DVL), pressure or compass) that satisfy the EKF assumptions. The article includes simulation results to illustrate the performance of this Near-Real-Time approach.