Autonomous Underwater Vehicle Docking 16.6 AUV Control Considerations 401
Part B | 16.6
proached from any direction, like pole docks. Slightly
more sophisticated homing control is required when
the dock must be approached from a particular direction, as is the case for a funnel-type dock. Control becomes more demanding for the directional
approach when there is a significant cross-current.
Even more complex homing is required in circumstances where there are obstacles, and soft docking is
required.
Pure pursuit, in which the vehicle always points
directly at the dock, is the simplest homing method.
In the event of a cross-current, pure pursuit results in
a hooked trajectory, in which the vehicle is blown to
one side of the dock. This can result in the vehicle missing the dock, and also means that the final approach
direction depends on the speed and direction of the current. This makes pure pursuit unattractive for docking
systems that require the direction of approach to be controlled. A variation on pure pursuit is to estimate the
current and use the current estimate to compute a heading correction that cancels the cross-current component
(the current orthogonal to the direction to the dock).
Wu et al. [16.54] described simulations of docking approaches in currents. This removes the hook from the
approach trajectory. Two-phase approaches are possible as well, for example, using an initial line of sight
approach and converting to controlled orientation in
the terminal phase, as simulated by Jantapremjit and
Wilson [16.55]. A fuzzy-behavioral approach was proposed by Smith et al. [16.56] and White et al. [16.57]
that drove the vehicle towards a virtual funnel to arrive
at a docking station.
The ability to measure the current environment in
the vicinity of the dock can substantially improve docking skills. An AUV equipped with a Doppler velocity
log will be able to measure current by computing the
difference of dead reckoned speed and measured speed
over bottom. For example, Yakimenko et al. [16.58]
describe a trajectory generation approach that can incorporate currents in planning the approach to the
dock. Docking methods that are omnidirectional in nature, like the pole dock, can use knowledge of the
current to approach into the current so as to minimize cross-current and reduce approach speed. If the
current cannot be measured directly by the AUV,
an alternative approach is for the dock to be free to
rotate and align with the current. If the AUV can
sense or interrogate the dock orientation, the near-field
docking trajectory will always be directly up-current,
avoiding the necessity for a crab angle of approach
to compensate for a cross-current. While this necessitates a method to communicate the dock orientation to the vehicle, some homing techniques like the
electromagnetic homing method described by Feezor
et al. [16.6] carry the information with the homing
signal.
16.6.2 Managing State: AUV Control
From a control perspective, docking of the AUV can be
understood as a sequence of tasks that culminate with
the vehicle connected to and secured by the dock. Each
of these tasks has specific sensing and control challenges, and in some cases actions by the vehicle must
be coordinated with actions by the dock. Overall management of the process involves sequencing between
tasks, possibly communicating with the dock, detecting failures, and invoking contingency activities when
a failure occurs. To achieve the individual tasks, and
to sequence between tasks, many operational systems
employ some variant of behavior-based control [16.59,
60].
The building blocks of behavior-based control are
the individual behaviors, which are control laws that
can be combined to accomplish tasks, which in turn
can be conditionally connected to achieve more complex missions or respond to contingencies. Bellingham
and Leonard [16.59] describe a behavior-based architecture, illustrating an implementation of homing as
used for under-ice homing and recovery. Docking work
reported by McEwen et al. [16.2], Feezor et al. [16.6],
and Singh et al. [16.5] employed the same behaviorbased architecture. Successful docking depends on the
vehicle carrying out a sequence of activities such as
setting up for the approach, making an approach, and
perhaps activating actuators upon contacting the dock to
secure itself and make connections for communications
and power. Figure 16.6 illustrates the state transition
table for an AUV for both docking and undocking, as
adapted from McEwen.
16.6.3 Managing State:
Coordinating AUV and Dock
Docking involves the coordination of two distinct entities, the dock and the AUV. Consequently, successful
control strategies for docking must address both platforms and contend with the reality that communications
between undersea systems are unreliable. Some docking designs require significant interaction between the
dock and the AUV, and consequently each platform
must be aware of the state of the other system in order to execute the appropriate action. Docking systems
that demand coordinated activities on the part of both
the vehicle and the dock introduce failure modes associated with the communication and coordination process.
Consequently, contingencies for failures must encompass the entire system.
Part B | 16.6
proached from any direction, like pole docks. Slightly
more sophisticated homing control is required when
the dock must be approached from a particular direction, as is the case for a funnel-type dock. Control becomes more demanding for the directional
approach when there is a significant cross-current.
Even more complex homing is required in circumstances where there are obstacles, and soft docking is
required.
Pure pursuit, in which the vehicle always points
directly at the dock, is the simplest homing method.
In the event of a cross-current, pure pursuit results in
a hooked trajectory, in which the vehicle is blown to
one side of the dock. This can result in the vehicle missing the dock, and also means that the final approach
direction depends on the speed and direction of the current. This makes pure pursuit unattractive for docking
systems that require the direction of approach to be controlled. A variation on pure pursuit is to estimate the
current and use the current estimate to compute a heading correction that cancels the cross-current component
(the current orthogonal to the direction to the dock).
Wu et al. [16.54] described simulations of docking approaches in currents. This removes the hook from the
approach trajectory. Two-phase approaches are possible as well, for example, using an initial line of sight
approach and converting to controlled orientation in
the terminal phase, as simulated by Jantapremjit and
Wilson [16.55]. A fuzzy-behavioral approach was proposed by Smith et al. [16.56] and White et al. [16.57]
that drove the vehicle towards a virtual funnel to arrive
at a docking station.
The ability to measure the current environment in
the vicinity of the dock can substantially improve docking skills. An AUV equipped with a Doppler velocity
log will be able to measure current by computing the
difference of dead reckoned speed and measured speed
over bottom. For example, Yakimenko et al. [16.58]
describe a trajectory generation approach that can incorporate currents in planning the approach to the
dock. Docking methods that are omnidirectional in nature, like the pole dock, can use knowledge of the
current to approach into the current so as to minimize cross-current and reduce approach speed. If the
current cannot be measured directly by the AUV,
an alternative approach is for the dock to be free to
rotate and align with the current. If the AUV can
sense or interrogate the dock orientation, the near-field
docking trajectory will always be directly up-current,
avoiding the necessity for a crab angle of approach
to compensate for a cross-current. While this necessitates a method to communicate the dock orientation to the vehicle, some homing techniques like the
electromagnetic homing method described by Feezor
et al. [16.6] carry the information with the homing
signal.
16.6.2 Managing State: AUV Control
From a control perspective, docking of the AUV can be
understood as a sequence of tasks that culminate with
the vehicle connected to and secured by the dock. Each
of these tasks has specific sensing and control challenges, and in some cases actions by the vehicle must
be coordinated with actions by the dock. Overall management of the process involves sequencing between
tasks, possibly communicating with the dock, detecting failures, and invoking contingency activities when
a failure occurs. To achieve the individual tasks, and
to sequence between tasks, many operational systems
employ some variant of behavior-based control [16.59,
60].
The building blocks of behavior-based control are
the individual behaviors, which are control laws that
can be combined to accomplish tasks, which in turn
can be conditionally connected to achieve more complex missions or respond to contingencies. Bellingham
and Leonard [16.59] describe a behavior-based architecture, illustrating an implementation of homing as
used for under-ice homing and recovery. Docking work
reported by McEwen et al. [16.2], Feezor et al. [16.6],
and Singh et al. [16.5] employed the same behaviorbased architecture. Successful docking depends on the
vehicle carrying out a sequence of activities such as
setting up for the approach, making an approach, and
perhaps activating actuators upon contacting the dock to
secure itself and make connections for communications
and power. Figure 16.6 illustrates the state transition
table for an AUV for both docking and undocking, as
adapted from McEwen.
16.6.3 Managing State:
Coordinating AUV and Dock
Docking involves the coordination of two distinct entities, the dock and the AUV. Consequently, successful
control strategies for docking must address both platforms and contend with the reality that communications
between undersea systems are unreliable. Some docking designs require significant interaction between the
dock and the AUV, and consequently each platform
must be aware of the state of the other system in order to execute the appropriate action. Docking systems
that demand coordinated activities on the part of both
the vehicle and the dock introduce failure modes associated with the communication and coordination process.
Consequently, contingencies for failures must encompass the entire system.