Underwater Vehicle Manipulators 17.5 Coordinated Motion Control of Underwater Vehicle-Manipulator Systems 413
Part B | 17.5
17.5 Coordinated Motion Control of Underwater Vehicle-Manipulator
Systems
The vehicle’s movements affect its manipulator’s and
vice versa, unless the vehicle is completely stationary during the manipulation by holding or sitting on
an underwater structure. The effect of the manipulator’s motion on a large underwater vehicle main body
is negligible, compared to the effect of the vehicle’s
motion on the accuracy of the manipulator’s end-effector. For a small vehicle, the coupled effect of the
movements from the vehicle main body and the manipulator is substantial and must be considered in the
overall control system design. This issue can be addressed by the coordinated motion control of the vehicle
and the manipulator.
The overall underwater vehicle-manipulator system
could be seen as a multibody system with redundant
degrees-of-freedom. The redundancy can be used to
assign additional motion without impeding the end-effector’s performance, and the motions of the vehicle
and the manipulator can be coordinated by redundancy
resolution of the overall system with optimizing an index. Therefore, the coordinated motion control of the
vehicle-manipulator system could be advantageous by
having additional degrees-of-freedom in manipulation.
The coordinated motion control system should have
a planning module capable of generating trajectories
for the vehicle and the manipulator for given tasks and
a control module capable of generating control signals
for the vehicle and the manipulator to accurately track
the trajectories in the presence of coupled effects and
uncertainties.
Several researchers have studied modeling and coordinated motion control of underwater vehicle-manipulator systems and showed that the coordinated motion control approach could improve the accuracy of
the end-effector [17.12–15, 17–20, 43, 52–67]. Some of
these previous studies are described below.
Schjølberg and Fossen [17.17] derived the dynamic
equations of an underwater vehicle-manipulator system, and applied feedback linearization for transforming the nonlinear system dynamics into a linear system for controller design. Antonelli [17.52] proposed
an adaptive control scheme that keeps the advantage of
model-based adaptive control while limiting the computational load. The control algorithm was designed
as a modular structure by applying the virtual decomposition approach [17.68]. Han et al. [17.59] proposed
a performance index for redundancy resolution. Redundant DOFs of the vehicle and manipulator were
used to optimize the index by applying the gradient
projection method. In addition, the inverse optimal
nonlinear H-infinity control was developed to guarantee tracking performance under uncertainty caused
by modeling or identification errors, or unmodeled
dynamics.
One must note that most commercial UUVs with
manipulators are heavy-duty large ROVs. They do not
have coordinated motion controllers for the vehicle
and manipulator systems and their manipulator controllers do not automatically compensate for errors due
to the vehicle motion. It is partly because the current
state of the art in positioning sensors such as long
baseline (LBL) does not offer the accuracy required
for a high performance of the manipulator, especially
its end-effector. The vehicle and the manipulator may
also operate in different bandwidths, mainly due to
the difference in size and dynamic characteristics. The
actual position of the vehicle remains within the accuracy of the vehicle position sensor. For example,
the accuracy of the vehicle position sensor of the order of a meter could mean that the actual vehicle
keeps moving within a meter and yet the sensor reads
the same position value. Since the vehicle is the base
of the manipulator, the required accuracy of the endeffector cannot be achieved without accurate sensory
information about the vehicle position. However, recent
developments in integrated navigation sensor technologies show very promising results for more accurate
position control of the vehicle, which would improve
the precision and accuracy of the end-effector. Panish and Taylor [17.69] experimentally demonstrated
that both the Kearfott T-24 and the IXSEA PHINS
III [17.70] provide exceptional navigation accuracies,
exceeding the published specifications–position drift of
less than 0:1% of distance travelled, CEP that refers to
the circular error probability, or a circle about a mean
value which includes 50% of the population. A drift
of 0:1% of distance travelled CEP means that the
position drift of less than 1 m for travelling 1 km between global positioning system (GPS) fixes is expected
for 50% of the time. Such developments could make
the practical use of the coordinated motion control
possible.
Part B | 17.5
17.5 Coordinated Motion Control of Underwater Vehicle-Manipulator
Systems
The vehicle’s movements affect its manipulator’s and
vice versa, unless the vehicle is completely stationary during the manipulation by holding or sitting on
an underwater structure. The effect of the manipulator’s motion on a large underwater vehicle main body
is negligible, compared to the effect of the vehicle’s
motion on the accuracy of the manipulator’s end-effector. For a small vehicle, the coupled effect of the
movements from the vehicle main body and the manipulator is substantial and must be considered in the
overall control system design. This issue can be addressed by the coordinated motion control of the vehicle
and the manipulator.
The overall underwater vehicle-manipulator system
could be seen as a multibody system with redundant
degrees-of-freedom. The redundancy can be used to
assign additional motion without impeding the end-effector’s performance, and the motions of the vehicle
and the manipulator can be coordinated by redundancy
resolution of the overall system with optimizing an index. Therefore, the coordinated motion control of the
vehicle-manipulator system could be advantageous by
having additional degrees-of-freedom in manipulation.
The coordinated motion control system should have
a planning module capable of generating trajectories
for the vehicle and the manipulator for given tasks and
a control module capable of generating control signals
for the vehicle and the manipulator to accurately track
the trajectories in the presence of coupled effects and
uncertainties.
Several researchers have studied modeling and coordinated motion control of underwater vehicle-manipulator systems and showed that the coordinated motion control approach could improve the accuracy of
the end-effector [17.12–15, 17–20, 43, 52–67]. Some of
these previous studies are described below.
Schjølberg and Fossen [17.17] derived the dynamic
equations of an underwater vehicle-manipulator system, and applied feedback linearization for transforming the nonlinear system dynamics into a linear system for controller design. Antonelli [17.52] proposed
an adaptive control scheme that keeps the advantage of
model-based adaptive control while limiting the computational load. The control algorithm was designed
as a modular structure by applying the virtual decomposition approach [17.68]. Han et al. [17.59] proposed
a performance index for redundancy resolution. Redundant DOFs of the vehicle and manipulator were
used to optimize the index by applying the gradient
projection method. In addition, the inverse optimal
nonlinear H-infinity control was developed to guarantee tracking performance under uncertainty caused
by modeling or identification errors, or unmodeled
dynamics.
One must note that most commercial UUVs with
manipulators are heavy-duty large ROVs. They do not
have coordinated motion controllers for the vehicle
and manipulator systems and their manipulator controllers do not automatically compensate for errors due
to the vehicle motion. It is partly because the current
state of the art in positioning sensors such as long
baseline (LBL) does not offer the accuracy required
for a high performance of the manipulator, especially
its end-effector. The vehicle and the manipulator may
also operate in different bandwidths, mainly due to
the difference in size and dynamic characteristics. The
actual position of the vehicle remains within the accuracy of the vehicle position sensor. For example,
the accuracy of the vehicle position sensor of the order of a meter could mean that the actual vehicle
keeps moving within a meter and yet the sensor reads
the same position value. Since the vehicle is the base
of the manipulator, the required accuracy of the endeffector cannot be achieved without accurate sensory
information about the vehicle position. However, recent
developments in integrated navigation sensor technologies show very promising results for more accurate
position control of the vehicle, which would improve
the precision and accuracy of the end-effector. Panish and Taylor [17.69] experimentally demonstrated
that both the Kearfott T-24 and the IXSEA PHINS
III [17.70] provide exceptional navigation accuracies,
exceeding the published specifications–position drift of
less than 0:1% of distance travelled, CEP that refers to
the circular error probability, or a circle about a mean
value which includes 50% of the population. A drift
of 0:1% of distance travelled CEP means that the
position drift of less than 1 m for travelling 1 km between global positioning system (GPS) fixes is expected
for 50% of the time. Such developments could make
the practical use of the coordinated motion control
possible.