Identification of a low-complexity flow field model for AUV applications

This paper presents identification algorithms for a low-dimensional planar flow model composed of a uniform flow component and a singular flow component. With the aim of improving navigation for small autonomous underwater vehicles operating in shallow water, we develop algorithms that can be performed in near real-time using small, sparse data sets. Although we compare several variations, the basic approach involves identifying the uniform flow component, the location of an assumed singularity, and the parameters which characterize the singular flow. In one variation, the flow model parameters are simply chosen to minimize a square error measure. In another variation, the minimization process is constrained to preserve the average divergence and circulation of the measured flow. The latter approach predictably results in a larger error, but produces a flow model which more realistically captures the flow in the larger region of the measurement points. Initial experimental results are presented