Autonomous underwater vehicles (AUVs) have been increasingly applied to oceanographic missions. Despite technological advances, adaptability to environmental changes or intelligent behaviour is still at an early stage of development, especially for large survey class AUVs. This research addresses the real-world challenge of using an AUV to delineate an oil plume in sea water, in the management of an environmental spill. A review of current literature showed the most common methodology in conventional adaptive AUV missions involved gradient-following methods with point-based sensors such as fluorometers. In this work, an adaptive sampling algorithm was developed for real discontinuous oil plumes and validated through simulation, The search and detection phase of the algorithm was tested through ocean trials with an Explorer survey-class AUV. In order to develop the algorithm, I complete a series of preparatory tests in a seawater tank, the ocean and in a lake. I used a seawater tank to assess the formation of oil in seawater, waves and currents and to assess the utility of sonar as a sensor for these oil droplets. The ocean was investigated to try to find an environment to test an AUV in an area of naturally occurring oil seeps. Finally, a lake was used to assess different proxies to represent oil in the ocean in order to avoid environmental damage and to obtain sonar records for use in developing real time analysis methods. The oil formation tests showed that real oil plumes are composed of countless undissolved droplets, especially after exposure to waves and currents. This result focused attention on the critical need for a sensor that could identify oil plumes at distance and a sampling strategy that could take into account the coalescent and clustering characteristics of real oil droplets in the water column. The seawater tank experiment found that a higher frequency sonar (1.35 MHz) was more effective than a lower frequency sonar (450 kHz) in capturing an oil plume in the water column. Thereafter, a scanning sonar (750 kHz frequency) was proposed as the primary sensor on the AUV in searching for plumes. An adaptive sampling mission can be achieved by utilising real-time processing of in-situ data to optimise an AUVs capability to complete a task, Here I developed modular algorithms for acoustic detection and in-situ analysis. In accordance with the use of sonar as an oil detection sensor, I introduced a new search path inspired by bumblebee flight patterns and validated this through field trials. This search design significantly improved the more conventional survey plans based on lawnmower paths by reducing the time to seek an oil plume by approximately 75.3%. For the detection strategy, the key assumption was to approximate an oil plume in a two-dimensional sensing layer with limited vertical extent. This improved detection performance by using a higher order two-dimensional scanning method that minimised the plausible impacts of mixing energies on the plume. A 'Measure ‚Äö- Analysis ‚Äö- Action' schema was employed in a recursive algorithm for oil plume tracking. The tracking modules analysed the recovered sensor data in real time to produce the desired best new heading of the AUV. To produce the best tracking performance, all the relevant parameters of each component term of the algorithm (such as thresholds, coefficients and weighting factor) had to be tuned and adjusted. Thereafter, I was able to track different forms of simulated oil plumes successfully with the AUV. The search and detection algorithms, as well as the backseat driver control system of the AUV using MOOS IvP routines (some of which I developed), were tested on the Explorer AUV during in ocean trials in Holyrood Bay, Newfoundland and Labrador, Canada. The algorithm implemented on the real AUV successfully demonstrated the AUV backseat driver control. Having appropriate configuration parameters, the guided path could be followed through updating the AUV state in pose (translational position (x, y) and rotational (˜í‚àè) angle). AUV heading direction was corrected by adapting to the computed desired heading, which was achieved in 2.8 seconds on average after the command was generated. Overall, the vehicle trajectory remained within 5.5m of the guided path. The field experiments verified the feasibility and utility of the designed search and detection algorithm in the ocean environment. Overall, this research has overturned the most common gradient-following methods employed in AUV-based oil spill investigation and tracking. The adaptive sampling system developed in this thesis provides an efficient tool to i) seek a realistic oil plume whose chemical characteristics may have altered over time, ii) detect patches of the plume with varying shapes and density and iii) track and map a plume without losing contact with it. The knowledge obtained through this research and the presented algorithm modules will support the next generation of researchers seeking to advance the autonomy of AUVs.
Copyright 2021 the author Chapter 2 appears to be the equivalent of a pre-print version of an article published as: Hwang, J., Bose, N., Fan, S., 2019. AUV adaptive sampling methods: A review, Applied sciences, 9(15), 3145. Copyright 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International (CC BY 4.0) license (http://creativecommons.org/licenses/by/4.0/). Chapter 5 appears to be the equivalent of a pre-print version of an article published as: Hwang, J., Bose, N., Nguyen, H. D., Williams, G., 2020. Acoustic search and detection of oil plumes using an autonomous underwater vehicle, Journal of marine science and engineering, 8(8), 618. Copyright 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International (CC BY 4.0) license (http://creativecommons.org/licenses/by/4.0/). Chapter 5 includes portions from the following published article: Hwang, J., Bose, N., Nguyen, H. D., Williams, G., 2020. Oil plume mapping: adaptive tracking and adaptive sampling from an autonomous underwater vehicle. IEEE Access, 8, 198021-198034. Copyright 2020. This work is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) License (https://creativecommons.org/licenses/by/4.0/).