Part B | 16.3
394 Part B Autonomous Ocean Vehicles, Subsystems and Control
scheme with quadrant detectors to reject uniform lighting [16.7]. In tests in San Diego Bay, when water
clarity reportedly obscured the yellow Odyssey AUV
submerged at depth of 2 m, the homing system was only
confused when submerged very shallowly (about 1 m or
less) and pointing towards a setting sun.
The use of multiple lights on the dock and a vision
system on the AUV allows more sophisticated homing.
With the appropriate configuration of lights, the vehicle can compute the vehicle position in dock coordinates. This is particularly useful if the dock is a cone or
some other system that requires the vehicle to approach
from a particular direction. Hong et al. [16.40] proposed
a system using five lights arranged around the perimeter
of a docking cone, and a sixth light in the same plane as
the other five, but outside of the cone. This arrangement
allowed the vehicle to compute its position and orientation relative to the dock when all dock lights were in the
field of view of the AUV imaging systems.
The five-light system implemented by Park et al.
was tested in a Korea Ocean Research and Development Institute (KORDI) engineering basin [16.41].
Park eliminated the sixth light, focusing on extracting
a directional cue and a range measurement. The environment provided challenges such as interfering light
sources, including reflection of the dock lights off the
water surface. Docking was initiated with the ISiMI
AUV [16.42] at a distance of 15 m pointing at the dock.
The authors report successful docking without using
range estimation, but also provide a range of suggestions for improving the image processing and control
algorithms.
Murarka et al. [16.20] used light-source homing for
recovery of their Endurance AUV through ice holes in
the Antarctic. This system used a flashing light, reminiscent of Cowen et al. [16.7], to guide the AUV to the
ice hole. However, the AUV was equipped with a vision
system rather than a simple quadrant detector. The light
source was suspended to the bottom of the ice hole,
where it would be detected by an upward looking camera on the AUV. Endurance is capable of hovering and
moving vertically. Once the flashing optical target was
detected, Endurance navigated to the ice hole entrance,
and then ascended through the 2 m-diameter, 3 m high
ice hole. The vision base systems accomplished both
approach to and ascent through the ice hole. The authors report using the algorithms for eight descents and
ten ascents through the ice.
Homing on Patterns and Arbitrary Structures
Vision systems can also be used to home on passive targets. These targets may be patterned to have
specific visual signatures useful to the vision processing algorithm. For example, Maire et al. [16.17] tested
a vision-based system for homing on black and white
striped poles. The system depended on ambient light
for illumination, and was tested in an outside diving
pool with the Starbug AUV. The vision processing algorithms were able to recognize and distinguish between
two different poles by the differing width of the black
and white stripes. The docking concept in this case consisted of landing on the bottom between the vertically
oriented poles.
Kushnerik et al. [16.43] describe a vision-based
docking system using a passive visual marker on the
bottom as the homing target. They explored a number of
potential marker patterns, finally selecting nested rectangles, as they provided orientation information. They
report using the tunnel sea lion (TSL) autonomous vehicle to demonstrate detection, servoing, and landing
on the markers using a down-looking camera in natural light at a water depth of 8 m in Peter the Great Bay.
16.3.3 Electromagnetic Homing Systems
Electromagnetic methods face a special challenge in
the ocean environment due to the conductive nature of
seawater. Not only are radio frequencies strongly attenuated, but the effects are frequency dependent. Below
gigahertz frequencies, attenuation goes as the inverse
of the root of the frequency. At a frequency of 1 MHz,
the 1=e attenuation distance, also called the skin depth,
is 0:3 m. However, at 1 kHz the skin depth increases
to approximately 10 m. Consequently, homing systems
using electromagnetic techniques must function at low
frequencies. Although the speed of propagation of electromagnetic signals is greatly slowed in a conducting
medium, it is still effectively instantaneous in the undersea environment. Thus the latency of an electromagnetic homing system is largely determined by the
implementation of the homing system, rather than the
physics of electromagnetic propagation in the ocean.
Magnetic Dipole Systems
Feezor et al. [16.6] developed an electromagnetic homing system for docking, using an oscillating magnetic
dipole to generate a homing signal. Since magnetic
fields are vector quantities, a sensor that fully resolves
magnetic components can determine the direction of the
magnetic field. By aligning the homing dipole with the
docking system, the magnetic measurement, in effect,
conveys the orientation of the dock. Thus the magnetic dipole system is attractive for docks that require
a directional approach, but whose orientation might be
unknown.
The oscillating nature of the magnetic homing signal creates a problem in that while the alignment of the
magnetic field can be determined, its absolute direction
394 Part B Autonomous Ocean Vehicles, Subsystems and Control
scheme with quadrant detectors to reject uniform lighting [16.7]. In tests in San Diego Bay, when water
clarity reportedly obscured the yellow Odyssey AUV
submerged at depth of 2 m, the homing system was only
confused when submerged very shallowly (about 1 m or
less) and pointing towards a setting sun.
The use of multiple lights on the dock and a vision
system on the AUV allows more sophisticated homing.
With the appropriate configuration of lights, the vehicle can compute the vehicle position in dock coordinates. This is particularly useful if the dock is a cone or
some other system that requires the vehicle to approach
from a particular direction. Hong et al. [16.40] proposed
a system using five lights arranged around the perimeter
of a docking cone, and a sixth light in the same plane as
the other five, but outside of the cone. This arrangement
allowed the vehicle to compute its position and orientation relative to the dock when all dock lights were in the
field of view of the AUV imaging systems.
The five-light system implemented by Park et al.
was tested in a Korea Ocean Research and Development Institute (KORDI) engineering basin [16.41].
Park eliminated the sixth light, focusing on extracting
a directional cue and a range measurement. The environment provided challenges such as interfering light
sources, including reflection of the dock lights off the
water surface. Docking was initiated with the ISiMI
AUV [16.42] at a distance of 15 m pointing at the dock.
The authors report successful docking without using
range estimation, but also provide a range of suggestions for improving the image processing and control
algorithms.
Murarka et al. [16.20] used light-source homing for
recovery of their Endurance AUV through ice holes in
the Antarctic. This system used a flashing light, reminiscent of Cowen et al. [16.7], to guide the AUV to the
ice hole. However, the AUV was equipped with a vision
system rather than a simple quadrant detector. The light
source was suspended to the bottom of the ice hole,
where it would be detected by an upward looking camera on the AUV. Endurance is capable of hovering and
moving vertically. Once the flashing optical target was
detected, Endurance navigated to the ice hole entrance,
and then ascended through the 2 m-diameter, 3 m high
ice hole. The vision base systems accomplished both
approach to and ascent through the ice hole. The authors report using the algorithms for eight descents and
ten ascents through the ice.
Homing on Patterns and Arbitrary Structures
Vision systems can also be used to home on passive targets. These targets may be patterned to have
specific visual signatures useful to the vision processing algorithm. For example, Maire et al. [16.17] tested
a vision-based system for homing on black and white
striped poles. The system depended on ambient light
for illumination, and was tested in an outside diving
pool with the Starbug AUV. The vision processing algorithms were able to recognize and distinguish between
two different poles by the differing width of the black
and white stripes. The docking concept in this case consisted of landing on the bottom between the vertically
oriented poles.
Kushnerik et al. [16.43] describe a vision-based
docking system using a passive visual marker on the
bottom as the homing target. They explored a number of
potential marker patterns, finally selecting nested rectangles, as they provided orientation information. They
report using the tunnel sea lion (TSL) autonomous vehicle to demonstrate detection, servoing, and landing
on the markers using a down-looking camera in natural light at a water depth of 8 m in Peter the Great Bay.
16.3.3 Electromagnetic Homing Systems
Electromagnetic methods face a special challenge in
the ocean environment due to the conductive nature of
seawater. Not only are radio frequencies strongly attenuated, but the effects are frequency dependent. Below
gigahertz frequencies, attenuation goes as the inverse
of the root of the frequency. At a frequency of 1 MHz,
the 1=e attenuation distance, also called the skin depth,
is 0:3 m. However, at 1 kHz the skin depth increases
to approximately 10 m. Consequently, homing systems
using electromagnetic techniques must function at low
frequencies. Although the speed of propagation of electromagnetic signals is greatly slowed in a conducting
medium, it is still effectively instantaneous in the undersea environment. Thus the latency of an electromagnetic homing system is largely determined by the
implementation of the homing system, rather than the
physics of electromagnetic propagation in the ocean.
Magnetic Dipole Systems
Feezor et al. [16.6] developed an electromagnetic homing system for docking, using an oscillating magnetic
dipole to generate a homing signal. Since magnetic
fields are vector quantities, a sensor that fully resolves
magnetic components can determine the direction of the
magnetic field. By aligning the homing dipole with the
docking system, the magnetic measurement, in effect,
conveys the orientation of the dock. Thus the magnetic dipole system is attractive for docks that require
a directional approach, but whose orientation might be
unknown.
The oscillating nature of the magnetic homing signal creates a problem in that while the alignment of the
magnetic field can be determined, its absolute direction