LINK FELLOWSHIP AWARDEES FOR 2010
OCEAN ENGINEERING AND INSTRUMENTATION


M. Jordan Stanway
Deep Submergence Laboratory
Massachusetts Institute of Technology and Woods Hole Oceanographic Institution
Joint Program in Oceanography / Applied Ocean Science and Engineering
Research Project: Advanced Underwater Vehicle Navigation
Advisor: Dr. Dana Yoerger, Senior Scientist, WHOI

Underwater vehicles are necessary to explore and study the ocean depths. Precise navigation increases the value of scientific data collected by these vehicles, and robust navigation enables autonomous operation in a variety of challenging conditions.

My research goal is to demonstrate a state estimator that provides robust precision navigation. The estimator is based on the sigma point Kalman filter framework. It supports fully coupled nonlinear models of vehicle dynamics, and provides mathematically rigorous treatment of delayed measurements. These two points are particularly important for underwater vehicles operating in a dynamic environment and without access to GPS.

Another goal is to demonstrate dead reckoning relative to an unknown ocean current profile using an acoustic doppler current profiler. This capability will allow underwater vehicles to maintain high-rate navigation during the descent and ascent phases of their missions. Sensor alignment is very important in precision navigation--I will use geometric algebra to formulate a novel alignment estimator, and compare it to currently available methods.

Each of these threads addresses a specific existing problem in underwater navigation. This research will provide novel approaches to solve each of these problems, and move the current state of the art toward more robust, higher precision navigation. This goal directly affects the value of all types of data collected by AUVs, ROVs, and human-occupied submersibles.


Gabrielle Inglis
Department of Ocean Engineering
University of Rhode Island
Research Project: A Lagrangian Imaging System for Sea-floor Mapping
Advisor: Dr. Chris Roman, Assistant Professor, Graduate School of Oceanography

High resolution sea floor imagery is valuable for fisheries management, habitat classification, ecosystem monitoring and estimation of geoacoustic seafloor properties. Imagery can be obtained using tethered drop cameras, towsleds, and robotic vehicles. However, each has specific operating constraints, costs, personnel requirements and risks.

Tethered drop cameras and small ROVs are relatively inexpensive imaging platforms, but have low stability and difficulty maintaining constant altitude. Low-cost towsleds can slide along the bottom to maintain a constant imaging distance but cannot operate in rocky or complex terrain. Additionally, each of these methods requires constant tether management and a dedicated support vessel. Autonomous Underwater Vehicles (AUVs) are free swimming imaging platforms that can fly structured surveys while maintaining a constant altitude in complex terrain. Widespread AUV use, however, is limited by the high upfront costs (typically ~100K and greater), peripheral costs and necessary operator expertise.

For this project, I will develop a novel concept in seafloor imaging by combining a new auto-ballasting terrain following Lagrangian float with a high resolution stereo camera system. The Lagrangian imaging platform will be a unique and useful instrument for many important problems in marine science, and be distinct from the current suite of underwater imaging tools.

Although not currently used for imaging, Lagrangian floats have been used for decades as sensing platforms in the deep ocean where they can drift and profile for years unsupervised. More recently, we have developed auto-ballasting floats for shallow water use, which can drift at a constant height above the bottom for several days in areas of varying bathymetry. The proposed instrument will be a Lagrangian camera system designed for multi-hour and multi-day missions. Its development is motivated by the desire to retain the desirable stability and constant altitude imaging characteristics of an AUV at substantially reduced cost.

Unlike an AUV, a float lacks horizontal propulsion and will drift with the currents. However, local drift track estimates still can be derived from periodic GPS locations and algorithmic means such as particle filter methodologies already used with AUVs. With knowledge of local currents, reasonably structured surveys will be achievable and provide adequate coverage when precise positioning is not required by the scientific objective. The resulting reduction in cost (target float cost ~15K), complexity and support means that many more instruments could be deployed by less skilled users.

Fully developing this concept will raise a number of interesting engineering problems related to real time field operation and subsequent data processing:

  • Real time adaptive frame rates to optimize coverage and power consumption
  • In-situ exposure setting and optimal camera/stobe separation
  • Multi-altitude imaging for scene representations at several scales
  • Image registration with existing structured optical surveys
  • Image derived current speed measurements

I will integrate existing systems and create a working prototype and then evaluate the above points through testing and data processing in order to identify and begin addressing the directions which most impact the further development of this concept.