Feature based adaptive energy management of sensors on autonomous underwater vehicles

Abstract

Autonomous Underwater Vehicles (AUVs) are an indispensable tool for studying the world׳s oceans. They can be equipped with sophisticated sensors and are capable of collecting scientific data under difficult and harsh conditions. Most AUVs rely on batteries to operate, making energy a critical resource. This paper evaluates the benefits of adaptive sampling strategies that allow AUVs to change sensing behaviors by modeling and tracking underwater features. For many AUVs, particularly gliders, these strategies can effectively extend mission duration. Often, necessary or likely preconditions of environmental features can be detected with simple, low energy cost sensors in combination with modeling software. Once precursors have been identified, more sophisticated and energy-hungry sensors can be utilized, collecting scientific data of higher quality. The use of both hardware and software trigger chains can reduce the energy consumption of sensors without sacrificing the collection of relevant data. We used thermocline detection software together with a lower powered sensor to trigger more expensive fluorescence and backscatter sensors on the Slocum Glider. The use of several different thermocline detection algorithms enabled us to compare their precision and power/energy savings. We show energy savings of up to 82% in simulations, and 50% and 63% through two at-sea experiments.