Autonomous Underwater Vehicle Docking 16.3 Sensors For Homing 393
Part B | 16.3
methods, and, consequently, array sizes must be significantly larger to get usable accuracy. For this reason,
homing methods using SBL methods have been implemented on docks rather than on vehicles.
An inverted SBL scheme was implemented by
Cowan [16.36] in which three radially distributed hydrophones on a dock measure the azimuth of a pulse
arriving from the vehicle. The resulting measurement
is encoded in two reply pulses, separated in frequency
and time. The time separation of the two pulses communicates the azimuth information. The attractive aspect
of this system is that it minimizes the complexity of
the system on the vehicle, reducing the job of the vehicle to simply measuring time-of-arrival of the two
pulses. However, some small latency is introduced by
the round-trip time. Latency is minimized both by making the system highly integrated and by operating it at
short ranges. Cowan’s system was intended to be used
at ranges of approximately 300 m or less.
Another inverted SBL system was tested by Smith
and Kronen [16.37] for application to docking. The
commercial Desert Star Dive Tracker system introduced sufficient latency, even at short ranges; Smith
concluded that the approach was unattractive for vehicles making a fast approach (a meter per second or
more). However, the approach was deemed useful for
slower moving vehicles.
Acoustic Imaging as a Docking Aid
Acoustics can be used to map undersea structures, thus
providing a potential cue for docking. The SWIMMER
vehicle, which is an AUV that delivers and connects
an ROV to subsea infrastructure, used an acoustic
imaging system for the final approach of the SWIMMER vehicle to a dock [16.38]. The preliminary tests
used an imaging system consisting of two orthogonally
mounted scanning sonars. The target consisted of the
docking station of which the vehicle was provided with
the CAD (computer-aided design) model. The imaging
system compared the actual sonar return to a simulation of the sonar return, generated with an estimated
position, the CAD model, and a sonar model, to generate updated positions. The system was demonstrated
with a mockup of the docking system in shallow water
at Cybemetix’ Marseilles harbor facilities.
16.3.2 Optical Homing
The advanced state of light sources, sensors, imaging
systems, image processing algorithms, computation,
and associated technologies provides a strong foundation for a variety of optical homing methods. Optical
systems can be roughly divided into some simple categories: systems that home on active lighting systems,
systems that home on structured patterns, and systems
capable of homing on arbitrary objects. The complexity
of the different approaches varies tremendously. Homing on an active light source can be one of the simplest
homing systems available, while software intensive
vision-based homing methodologies require substantial
computational capability, and even then may be limited
in update rate.
The acquisition range of optical systems is, in general, greatly limited compared to acoustics in natural
ocean environments. Attenuation of optical signals occurs due to scattering and absorption of light by both
seawater itself, and the particulate and dissolved constituents of seawater [16.39]. The optical properties
of natural seawater are highly wavelength dependent,
with the best clarity coinciding roughly with those of
human vision. The presence of particulates and dissolved chemicals greatly changes the absorption and
scattering properties of water, and can increase attenuation many order of magnitude. As a consequence, light
propagation in the ocean is highly location and depthdependent. Proximity to coastal runoff, phytoplankton
blooms, seafloor sediment, and other sources of particles and dissolved compounds can dramatically change
optical properties and, therefore, the performance of optical homing systems.
Homing on Active Light Sources
Cowen et al. [16.7, 36] demonstrated an elegant optical homing system using a chopped light source as the
homing target and a quadrant detector on the AUV to
provide homing cues. The system provided right–left
and up–down cues to the vehicle. The directional nature of both the light source and the detector meant
that the vehicle needed to approach the dock towards
the light source in order to detect the source. Despite
water clarity so poor that the vehicle was not visible
at a depth of 2 m, acquisition distances of the optical
system were 1015 m. The performance of the optical
homing system in Buzzard Bay demonstrated acquisition distances of roughly 28 m. The author assisted with
these tests and observed that when the Odyssey AUV
acquired lock on the light source at sufficient distance
(more than a few meters), it always hit the cone dock
target.
Ambient illumination from the sun or other light
sources poses a potential source of confusion when
homing on the light source. The system described by
Cowen discriminated against ambient light using two
techniques. By modulating the light source at 40 Hz
the system could reject the more slowly varying ambient light using a high-pass filter. The more spatially
constant ambient light could also be rejected, as compared to the point source target, by using a differencing
Part B | 16.3
methods, and, consequently, array sizes must be significantly larger to get usable accuracy. For this reason,
homing methods using SBL methods have been implemented on docks rather than on vehicles.
An inverted SBL scheme was implemented by
Cowan [16.36] in which three radially distributed hydrophones on a dock measure the azimuth of a pulse
arriving from the vehicle. The resulting measurement
is encoded in two reply pulses, separated in frequency
and time. The time separation of the two pulses communicates the azimuth information. The attractive aspect
of this system is that it minimizes the complexity of
the system on the vehicle, reducing the job of the vehicle to simply measuring time-of-arrival of the two
pulses. However, some small latency is introduced by
the round-trip time. Latency is minimized both by making the system highly integrated and by operating it at
short ranges. Cowan’s system was intended to be used
at ranges of approximately 300 m or less.
Another inverted SBL system was tested by Smith
and Kronen [16.37] for application to docking. The
commercial Desert Star Dive Tracker system introduced sufficient latency, even at short ranges; Smith
concluded that the approach was unattractive for vehicles making a fast approach (a meter per second or
more). However, the approach was deemed useful for
slower moving vehicles.
Acoustic Imaging as a Docking Aid
Acoustics can be used to map undersea structures, thus
providing a potential cue for docking. The SWIMMER
vehicle, which is an AUV that delivers and connects
an ROV to subsea infrastructure, used an acoustic
imaging system for the final approach of the SWIMMER vehicle to a dock [16.38]. The preliminary tests
used an imaging system consisting of two orthogonally
mounted scanning sonars. The target consisted of the
docking station of which the vehicle was provided with
the CAD (computer-aided design) model. The imaging
system compared the actual sonar return to a simulation of the sonar return, generated with an estimated
position, the CAD model, and a sonar model, to generate updated positions. The system was demonstrated
with a mockup of the docking system in shallow water
at Cybemetix’ Marseilles harbor facilities.
16.3.2 Optical Homing
The advanced state of light sources, sensors, imaging
systems, image processing algorithms, computation,
and associated technologies provides a strong foundation for a variety of optical homing methods. Optical
systems can be roughly divided into some simple categories: systems that home on active lighting systems,
systems that home on structured patterns, and systems
capable of homing on arbitrary objects. The complexity
of the different approaches varies tremendously. Homing on an active light source can be one of the simplest
homing systems available, while software intensive
vision-based homing methodologies require substantial
computational capability, and even then may be limited
in update rate.
The acquisition range of optical systems is, in general, greatly limited compared to acoustics in natural
ocean environments. Attenuation of optical signals occurs due to scattering and absorption of light by both
seawater itself, and the particulate and dissolved constituents of seawater [16.39]. The optical properties
of natural seawater are highly wavelength dependent,
with the best clarity coinciding roughly with those of
human vision. The presence of particulates and dissolved chemicals greatly changes the absorption and
scattering properties of water, and can increase attenuation many order of magnitude. As a consequence, light
propagation in the ocean is highly location and depthdependent. Proximity to coastal runoff, phytoplankton
blooms, seafloor sediment, and other sources of particles and dissolved compounds can dramatically change
optical properties and, therefore, the performance of optical homing systems.
Homing on Active Light Sources
Cowen et al. [16.7, 36] demonstrated an elegant optical homing system using a chopped light source as the
homing target and a quadrant detector on the AUV to
provide homing cues. The system provided right–left
and up–down cues to the vehicle. The directional nature of both the light source and the detector meant
that the vehicle needed to approach the dock towards
the light source in order to detect the source. Despite
water clarity so poor that the vehicle was not visible
at a depth of 2 m, acquisition distances of the optical
system were 1015 m. The performance of the optical
homing system in Buzzard Bay demonstrated acquisition distances of roughly 28 m. The author assisted with
these tests and observed that when the Odyssey AUV
acquired lock on the light source at sufficient distance
(more than a few meters), it always hit the cone dock
target.
Ambient illumination from the sun or other light
sources poses a potential source of confusion when
homing on the light source. The system described by
Cowen discriminated against ambient light using two
techniques. By modulating the light source at 40 Hz
the system could reject the more slowly varying ambient light using a high-pass filter. The more spatially
constant ambient light could also be rejected, as compared to the point source target, by using a differencing