Part B | 13.2
326 Part B Autonomous Ocean Vehicles, Subsystems and Control
13.2 Autonomous Maneuvering and Navigation
Most USVs today require substantial, remote human
assistance to maneuver and navigate in anything but
a very highly controlled environment. Consequently,
a reliable, high-bandwidth radio link to transmit imagery from the USV to the human operator for hazard
avoidance is required; they are not capable of autonomous operation. On the other hand, an autonomous
USV (AUSV) is capable of perceiving its internal and
external environment, capable of performing a task
or function in a complex environment with minimal
human intervention, and capable of responding effectively to a dynamic situation. A non-autonomous USV
includes control approaches such as remote-control,
waypoint navigation, and scripted operations, and has
no organic perception – it is not capable of sensing or
responding to its environment. In contrast, an AUSV is
capable of hazard avoidance in compliance with the International Regulations for Preventing Collisions at Sea
1972 (COLREGs) without human assistance, can arbitrate between multiple, competing objectives, and can
do planning/replanning based on changing mission objectives, situations, and resource levels.
The state-of-the-art for AUSVs is on-water demonstrations in good visibility with avoidance of fixed
hazards for own ship speeds of > 20 kn, avoidance of
moving hazards with COLREGs compliance for own
boat speeds of > 20 kn, implementation of COLREGs
for head-on, overtake, and crossing situations, and tactically relevant behaviors, including go-to-waypoint,
avoid hazard, obey COLREGs, trail, intercept, and patrol. Real-time adaptive response to vehicle faults and
dynamic re-planning have also been demonstrated onwater. An example of arbitration between multiple,
competing objectives is the autonomous maneuvering
and navigation of a USV in achieving a particular
waypoint while avoiding hazards and complying with
COLREGs.
The sea surface environment presents significant
challenges for autonomous maneuvering and navigation. Issues include intermittent obscuration of sensors
in rough seas by sea spray and wave slap, limitations of
radar close to the USV, the effect of platform motions
on sensor performance, the effect of sea surface topography on sensor line-of-sight, and limitations imposed
by atmospheric weather conditions (fog, rain, snow,
marine layer) on sensor performance. There is work underway to develop advanced autonomy for USVs and
address the above technical challenges [13.13–15]. This
is enabled by continuing advancements in technologies
such as sensors (EO (electro-optic), IR (infrared), radar,
LIDAR (light detection and ranging)), sensor processing, sensor fusion, perception capable of dynamically
adjusting to environmental conditions, determination of
intent (of other vessels), decision-making using uncertain and intermittent perception information, and lookahead planning. A highly dynamic and unpredictable
environment combined with mission complexity motivates a hybrid autonomy architecture [13.14–16] that
includes a reactive decision-making component and
a deliberative decision-making component. The reactive component provides a fast answer and is used for
quick response to close-aboard hazards not detected by
the perception system until the last moment and is necessary for safe operations in high risk environments.
Deliberative decision-making uses more information
and provides an answer that is valid over a longer period of time (unless the situation changes!), but it takes
longer for the system to provide that answer. Deliberative planning is an enabler for long-duration operations.
A hybrid autonomous architecture has been demonstrated on-water [13.14–16].
Widespread adoption of autonomous maneuvering and navigation will require that significant nontechnical challenges be overcome to properly address
safety/liability concerns. Yet to be defined are standard metrics of autonomy in terms of core competencies with associated proficiency tests. Approaches
consisting purely of on-water testing to a high level
of confidence in all conditions are prohibitively expensive; therefore, modeling and simulation will likely play
a key role. Cultural issues will be more difficult for
AUSVs than for unmanned systems. Non-autonomous
unmanned systems have had to overcome the issue
of taking the human out of the vehicle and locating
him/her remotely from the platform, but still having
a high degree of involvement with the operation of the
platform. AUSVs will confront the issue of the human being much less involved in the operation, with
the involvement being at a supervisory level only. Acceptance of this mode of operation will occur through
a gradual approach in which AUSVs are initially used
in an autonomous mode, but with a high degree of
human supervision to establish trust in the system’s performance. As confidence grows, the degree of human
oversight will be decreased, allowing AUSVs to reach
their full potential.
326 Part B Autonomous Ocean Vehicles, Subsystems and Control
13.2 Autonomous Maneuvering and Navigation
Most USVs today require substantial, remote human
assistance to maneuver and navigate in anything but
a very highly controlled environment. Consequently,
a reliable, high-bandwidth radio link to transmit imagery from the USV to the human operator for hazard
avoidance is required; they are not capable of autonomous operation. On the other hand, an autonomous
USV (AUSV) is capable of perceiving its internal and
external environment, capable of performing a task
or function in a complex environment with minimal
human intervention, and capable of responding effectively to a dynamic situation. A non-autonomous USV
includes control approaches such as remote-control,
waypoint navigation, and scripted operations, and has
no organic perception – it is not capable of sensing or
responding to its environment. In contrast, an AUSV is
capable of hazard avoidance in compliance with the International Regulations for Preventing Collisions at Sea
1972 (COLREGs) without human assistance, can arbitrate between multiple, competing objectives, and can
do planning/replanning based on changing mission objectives, situations, and resource levels.
The state-of-the-art for AUSVs is on-water demonstrations in good visibility with avoidance of fixed
hazards for own ship speeds of > 20 kn, avoidance of
moving hazards with COLREGs compliance for own
boat speeds of > 20 kn, implementation of COLREGs
for head-on, overtake, and crossing situations, and tactically relevant behaviors, including go-to-waypoint,
avoid hazard, obey COLREGs, trail, intercept, and patrol. Real-time adaptive response to vehicle faults and
dynamic re-planning have also been demonstrated onwater. An example of arbitration between multiple,
competing objectives is the autonomous maneuvering
and navigation of a USV in achieving a particular
waypoint while avoiding hazards and complying with
COLREGs.
The sea surface environment presents significant
challenges for autonomous maneuvering and navigation. Issues include intermittent obscuration of sensors
in rough seas by sea spray and wave slap, limitations of
radar close to the USV, the effect of platform motions
on sensor performance, the effect of sea surface topography on sensor line-of-sight, and limitations imposed
by atmospheric weather conditions (fog, rain, snow,
marine layer) on sensor performance. There is work underway to develop advanced autonomy for USVs and
address the above technical challenges [13.13–15]. This
is enabled by continuing advancements in technologies
such as sensors (EO (electro-optic), IR (infrared), radar,
LIDAR (light detection and ranging)), sensor processing, sensor fusion, perception capable of dynamically
adjusting to environmental conditions, determination of
intent (of other vessels), decision-making using uncertain and intermittent perception information, and lookahead planning. A highly dynamic and unpredictable
environment combined with mission complexity motivates a hybrid autonomy architecture [13.14–16] that
includes a reactive decision-making component and
a deliberative decision-making component. The reactive component provides a fast answer and is used for
quick response to close-aboard hazards not detected by
the perception system until the last moment and is necessary for safe operations in high risk environments.
Deliberative decision-making uses more information
and provides an answer that is valid over a longer period of time (unless the situation changes!), but it takes
longer for the system to provide that answer. Deliberative planning is an enabler for long-duration operations.
A hybrid autonomous architecture has been demonstrated on-water [13.14–16].
Widespread adoption of autonomous maneuvering and navigation will require that significant nontechnical challenges be overcome to properly address
safety/liability concerns. Yet to be defined are standard metrics of autonomy in terms of core competencies with associated proficiency tests. Approaches
consisting purely of on-water testing to a high level
of confidence in all conditions are prohibitively expensive; therefore, modeling and simulation will likely play
a key role. Cultural issues will be more difficult for
AUSVs than for unmanned systems. Non-autonomous
unmanned systems have had to overcome the issue
of taking the human out of the vehicle and locating
him/her remotely from the platform, but still having
a high degree of involvement with the operation of the
platform. AUSVs will confront the issue of the human being much less involved in the operation, with
the involvement being at a supervisory level only. Acceptance of this mode of operation will occur through
a gradual approach in which AUSVs are initially used
in an autonomous mode, but with a high degree of
human supervision to establish trust in the system’s performance. As confidence grows, the degree of human
oversight will be decreased, allowing AUSVs to reach
their full potential.