Adaptive estimator for automatic guidance of an unmanned submersible

Abstract

The automatic guidance of an unmanned submersible using an acoustic navigation system requires a state estimator that can predict the vehicle position during the navigation sample period, filter the random measurement disturbance and estimate the mean disturbance due to sea current. The hydrodynamic parameters of the vehicle are nonlinear and can vary with the magnitude and direction of the relative fluid velocity within the duration of the period of the navigation sample. A prediction-correction algorithm is described that incorporates real-time gain adaption to minimise the filter-error variance and a correction to offset the bias of the model and sea current disturbances.