A comparison of two approaches for adaptive sampling of environmental processes using autonomous underwater vehicles

Two classes of adaptive sampling of underwater processes are considered for autonomous underwater vehicle (AUV) applications. The first approach is based on estimating parameters of an assumed process model using either Kalman filter or least squares techniques. The second, nonparametric approach is based on information-theoretic concepts and incorporates a classification phase in lieu of a process model. Two applications of each method are evaluated for processes with closed boundaries. Specifically, we utilize a finite element simulation of neutral tracer injection advected by a turbulent flow field