Abstract

Ocean surveying is the acquisition of acoustic data representing various features of the seafloor and the water above it, including water depth, seafloor composition, the presence of fish, and more. Historically, this was a task performed solely by manned vessels, but with advances in robotics and sensor technology, autonomous surface vehicles (ASVs) with sonar equipment are beginning to supplement and replace their more costly manned counterparts. The popularity of these vessels calls for advances in software to control them.

In this thesis we define the problem of path coverage to represent and generalize that of ocean surveying, and propose a real-time motion planning algorithm to solve it. We prove theorems of completeness and local asymptotic optimality regarding the proposed algorithm, and evaluate it in a simulated environment. We also discover a lack of robustness in the Dubins vehicle model when applied to real-time motion planning. We implement a model-predictive controller and other components for an autonomous surveying system, and evaluate it in simulation. The system documented in this thesis takes a step towards fully autonomous ocean surveying, and proposes further extensions which get even closer to that goal.

Presenter Bio

Alex Brown graduated with a bachelor's degree in computer science in 2018 from the University of New Hampshire, and is pursuing a master's degree in computer science. Before joining the Center in May of 2019, he spent a year on the IT team at the UNH Interoperability Lab and a couple summers as a software development intern at the predictive analytics and data management company Rapid Insight. His research interests include planning, machine learning and artificial intelligence in general.

Publication Date

8-3-2020

Document Type

Presentation

Download

Share

COinS