19 / 34
ACT Autonomous Surface Vehicle Workshop Report
Page
19
of
34
payloads. In this configuration the manned vessel takes on the responsibility of both
reactionary and deliberative collision avoidance for itself and the autonomous vessels
under its care. The autonomous vessels might operate at an Operational Level 3
autonomy receiving real-time position, heading and speed of the parent ship via a
telemetry link and automatically adjusting mission parameters to maintain relative
position, or Operational Level 4 autonomy actively sensing the position of the mother
ship in real time via sensors such as radar or Lidar to do the same. Similarly, vessels
having Level 2 Self-awareness, i.e. the ability to monitor and log operating parameters
and to generate faults to warn operators operate at much lower risk than those that do
not. Preferable still, is Level 3 Self-awareness in which each vessel understands its own
turn radius and other characteristics and can anticipate maneuvers to keep in lock-step
with the manned vessel through turns and avoidance maneuvers. The autonomous
vessels will operate sensor payloads and while this is possible at a Level 1 (piloting) level
of sensor autonomy, the operation is more tractable at level’s 2, 3 or 4, giving the
sensors the ability to start and stop logging automatically and the ability for an operator
to configure the sensor during the mission or for it to reconfigure itself automatically to
optimize quality data collection.
Now consider a 20 ft. autonomous vessel conducting hydrographic survey with Level 4
Self-awareness (i.e. the ability to monitor internal systems, model their operation and
that of the vessel as a whole and to recognize when they do not match expectations),
Level 4 Operational autonomy (i.e. the ability to follow a mission plan, to sense
obstacles and avoid them) and Level 4 Sensor autonomy (the ability to adjust sensor
configurations to optimize data collection). When operating in distant waters such the
Alaskan coast, where other vessel traffic and obstacles are unlikely, this vessel might
operate securely in an independent mode of supervision. With systems in place to
monitor, log and alarm on the vehicle’s internal health and a model in place to verify the
system is responding to commands appropriately, operators could feel secure that
everything is working smoothly. Full COLREGS autonomy may not be in place, nor even
necessary in this environment, but some ability to recognize a major obstruction and
avoid it, or even simply to warn operators when such an obstruction exists will help
ensure the vessel does not collide with an unexpected obstacle. Moreover, smart
sensors capable of adjusting operating parameters as conditions change and processes
capable of assessing data quality and/or sending data review information to the
operator via low bandwidth telemetry link would ensure that data of high quality are
collected. The vessel may operate independently because the vessel’s systems closely
monitor themselves and because external hazards are both unlikely and actively sensed,
mitigating the risks involved.
Operate this same vessel in New York Harbor however, even with full COLREGS
compliant autonomy, and the level of risk involved might not warrant fully independent
operation. In this case the multitude of both static and dynamic obstacles combined
with the high visibility of the location might increase the level of risk beyond comfort
levels for most operators. One can imagine, what now seems futuristic, a scenario in
which sensor systems to detect other vessels and obstacles as well as algorithms to