A Virtual Ocean framework for environmentally adaptive, embedded acoustic navigation on autonomous underwater vehicles

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology and the Woods Hole Oceanographic Institution September 2021. Autonomous underwater vehicles (AUVs) are an increasingly capable robotic platform, with embedded acou...

Full description

Bibliographic Details
Main Author: Bhatt, EeShan C.
Other Authors: Schmidt. Henrik
Format: Thesis
Language:English
Published: Massachusetts Institute of Technology and Woods Hole Oceanographic Institution 2021
Subjects:
Online Access:https://hdl.handle.net/1912/27309
Description
Summary:Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology and the Woods Hole Oceanographic Institution September 2021. Autonomous underwater vehicles (AUVs) are an increasingly capable robotic platform, with embedded acoustic sensing to facilitate navigation, communication, and collaboration. The global positioning system (GPS), ubiquitous for air- and terrestrial-based drones, cannot position a submerged AUV. Current methods for acoustic underwater navigation employ a deterministic sound speed to convert recorded travel time into range. In acoustically complex propagation environments, however, accurate navigation is predicated on how the sound speed structure affects propagation. The Arctic’s Beaufort Gyre provides an excellent case study for this relationship via the Beaufort Lens, a recently observed influx of warm Pacific water that forms a widespread yet variable sound speed lens throughout the gyre. At short ranges, the lens intensifies multipath propagation and creates a dramatic shadow zone, deteriorating acoustic communication and navigation performance. The Arctic also poses the additional operational challenge of an ice-covered, GPSdenied environment. This dissertation demonstrates a framework for a physics-based, model-aided, real-time conversion of recorded travel time into range—the first of its kind—which was essential to the successful AUV deployment and recovery in the Beaufort Sea, in March 2020. There are three nominal steps. First, we investigate the spatio-temporal variability of the Beaufort Lens. Second, we design a human-in-the-loop graphical decision-making framework to encode desired sound speed profile information into a lightweight, digital acoustic message for onboard navigation and communication. Lastly, we embed a stochastic, ray-based prediction of the group velocity as a function of extrapolated source and receiver locations. This framework is further validated by transmissions among GPS-aided modem ...