Knowing the transponder location for Autonomous Underwater Vehicle (AUV) navigationis imperative to predicting the location of the AUV during mission navigation. Accurate
underwater navigation requires external transponder beacons or navigation aids with known
locations. Traditionally, the localization method for these transponder beacons is normally done
with ship-based surveys that take time and personnel. This thesis proposes that using autonomous
vehicles, specifically surface vehicles, to perform the transponder survey will free up personnel,
save time, and yield accurate and precise estimates of the transponder location. This thesis looks
to automate the process of transponder beacon navigation.
xi
Two approaches are applied and developed to perform localization: a least squares
method and an extended Kalman Filter. These approaches are tested on field data collected
by a Boeing Liquid Robotics Wave Glider equipped with a WHOI micromodem for acoustic
communications and a Global Positioning System (GPS) receiver. Two field experiments were
conducted in La Jolla, California in shallow water, and included varying geometries of the survey
path used for localization.
The Kalman Filter outperforms the least squares method for precision, and it identifies
the error bounds of the estimate. It includes a motion model specific to the physical movement
of the Wave Glider’s sub and float components. Finally, the Kalman filter provides an on-board
algorithm that can be run in real-time without excessive usage of data storage, which the least
squares method would require.
Both theoretical and data analysis conclude that traveling a 150-degree arc around the
drop location will allow the area of uncertainty and transponder position covariance error ellipse
to converge to a steady state value. Traveling straight line transects will also yield precise survey
results and may diminish the total time it takes to perform the localization survey.