Optical and Acoustic Perception for Robotic Exploration and Intervention in Challenging Underwater Environments
Name
phung-aphung-phd-aeroastro-2026-thesis.pdf
Size
46.49 MB
Format
Adobe PDF
Checksum (MD5)
bb38d0ce6a456ede979beb3afd1160bf
Author(s)
Phung, Amy Ngo
Advisor(s)
Camilli, Richard
Date Issued
May 2026
Publisher
Massachusetts Institute of Technology
Abstract
Underwater robotic systems are essential for conducting scientific exploration, construction, and routine maintenance tasks, and will play a crucial role in future missions to explore oceans of other planetary bodies beyond Earth. These systems are vital for collecting large-scale datasets which improve our understanding of natural environments, conducting georeferenced surveys in sensitive regions to track the local impact of human activity, and completing intervention tasks to maintain the offshore infrastructure people rely on globally for ubiquitous resources such as internet and power (e.g., oil and gas, offshore wind energy).
The conventional approach for recording large-scale datasets and surveys involves autonomous underwater vehicles (AUVs). Low-cost, person-portable AUVs only possess a basic degree of autonomy, and are typically limited to basic waypoint-to-waypoint missions for data collection in open waters. Larger AUVs carry more power and specialized sensors, and are capable of more complex autonomy (e.g., following a target organism, navigating near structures). However, their resource-intensive design requires significant ship-based infrastructure for deployment and recovery.
Subsea intervention tasks (e.g., scientific seafloor sampling, infrastructure maintenance) are performed by remotely operated vehicles (ROVs), which are directly teleoperated by a human pilot from a surface support vessel across a tethered connection. Ongoing research aims to reduce the substantial training and infrastructure requirements required for ROV operations by integrating various degrees of robot autonomy. However, due to their reliance on optical cameras for perception, both teleoperated and autonomy-based ROV operations can be significantly hampered by turbid conditions caused by natural or anthropogenic causes (e.g., wind and waves, marine traffic, and urban runoff).
Currently, the substantial infrastructure requirements of conventional AUVs and ROVs constrain operations to ship-accessible regions, increase operational costs, and necessitate strict coordination of vessel schedules due to the shared use of a limited number of assets. These extensive resource demands further hinder deployment in remote or operationally challenging environments. Future vehicle concepts envision distributed fleets that provide expanded coverage while requiring significantly less supporting infrastructure. However, the lack of sufficiently robust perception capabilities remains a critical bottleneck, limiting the level of autonomy necessary to fully realize this vision.
The work presented in this thesis aims to address the field’s growing need for robust onboard perception in operationally challenging environments. To this end, the thesis introduces: (1) a hybrid onboard control architecture that extends capabilities of resource-intensive AUVs to low-power vehicles; (2) an automated acoustic surface ice detection method; (3) acoustic perception methods that enable reliable 3D reconstruction in turbid conditions; (4) real-time, computationally efficient opti-acoustic fusion methods that leverage optical data to improve the semantic interpretation and resolution of acoustic reconstructions; and (5) methods for 3D opti-acoustic segmentation that enable object identification and manipulation in turbid conditions. Collectively, these contributions aim to enable safer and more effective
operations in undersampled environments such as turbid surf zones and Arctic regions, while supporting the transition toward future remote and autonomous vehicle concepts.
MIT Department
Massachusetts Institute of Technology. Department of Aeronautics and Astronautics
Joint Program in Oceanography/Applied Ocean Science and Engineering
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
