Control of a Relocatable Energy-Harvesting Autonomous Underwater Vehicle in a Spatiotemporally-Varying Gulf Stream Resource

This paper describes and evaluates, through data-driven simulation, a strategy for using a relocatable autonomous underwater vehicle (AUV) with on-board turbines to extract hydrokinetic energy out of a shifting Gulf Stream. The conceptual AUV generates energy while anchored, and can relocate itself when needed. In order to maintain an estimate of the Gulf Stream flow profile that can be referenced for control, an estimator blends a measurement of the flow speed at the AUV's location with intermittent secondary flow speed data (which is available through a High Frequency Radar Network (HFRNet)). Model predictive control (MPC)-based spatiotemporal optimization is then used to optimize the location of the AUV along a cross-stream transect whose velocity profile varies as a function of time and cross-stream location. The proposed MPC strategy is structured in a way that the need for instantaneous power maximization (termed exploitation) is balanced with the need to maintain an accurate map of ocean current speed versus location on transect (termed exploration). The control strategy is validated through data from a numerical model of coastal circulation dynamics.