Distributed Flow Estimation for Autonomous Underwater Robots Using POD-Based Model Reduction

This paper presents a novel flow estimation approach that assimilates distributed pressure measurements of autonomous underwater robots through coalescing recursive Bayesian estimation and proper orthogonal decomposition (POD)-based flow model reduction. The proposed flow estimation approach does not rely on any analytical flow models and is thus applicable to many and various complicated flow fields for arbitrarily shaped underwater robots while most of the existing flow estimation methods apply only to those with simple and well-defined shapes. Neural network is further used to establish the relationship between the POD flow model and the flow parameters of interest, e.g., the angle of attack and flow-relative velocity. To demonstrate the effectiveness of the proposed distributed flow estimation approach, two simulation studies, one with a circular-shaped robot and one with a Joukowski-foil-shaped robot, are presented.