16 / 34
ACT Autonomous Surface Vehicle Workshop Report
Page
16
of
34
warn operators or other vessels, when possible, compensate for the effect, all to
mitigate their impact on operations, risk to property or human life. Advanced Autonomy
involves closed loop control in which, for example, movements of the rudder are tested
against expected changes in heading, or increases in thrust are tested against changes in
vessel speed. Advanced Autonomy also includes the ability to recognize when one of a
redundant set of systems has failed and to switch operations to the second. Reactions
to faults also might include for example, sounding five blasts of a whistle in accordance
with International Regulations for Preventing Collisions at Sea 1972 (COLREGS) when a
loss of propulsion is detected. Advanced Autonomy is distinguished from Intermediate
Autonomy in that control actions are actively tested against a model of expected vehicle
response and action is taken when the two do not match.
Operations: Advanced Autonomy involves the ability to adjust a pre-planned mission in
a reactionary way to dynamically sensed conditions, for example, to detect and avoid
previously unknown buoys, lobster pot floats, other vessels (COLREGS compliance), to
follow and/or track another vessel and to moor by anchor or pier without user
intervention. Advanced Autonomy is distinguished from Intermediate Autonomy in that
it requires onboard sensing systems to actively measure dynamically changing
conditions and well defined behaviors to react to the input from those sensors.
Advanced Autonomy is distinguished from Planning in that the behaviors that result are
generally short-term changes to operations, while Planning involves longer-term
strategic changes to operation. Both may involve multiple competing objectives.
Sensor: Advanced Autonomy involves the ability to adjust sensor configurations in a
reactionary way to input from other sensors or mission conditions. For example, a
bathymetric sonar may increase transmit power levels when signal to noise ratio is
deemed to be too low to obtain high quality bottom detections or for example, a
camera’s shutter speed may be increased when lighting threatens to saturate images.
Advanced Autonomy is distinguished from Intermediate Autonomy in that it provides
the ability for sensors to adjust their configuration to optimal parameters without
operator interaction.
Level 5: Planning (think)
Self-awareness, Operations and Sensor: At the Planning Level of Autonomy, the three
categories that have been useful to keep separate thus far merge into one. Planning
involves the ability to make a major adjustment or totally create a pre-planned mission
based on a deliberative consideration of objectives, fuel/power physical constraints, and
both previously known fixed obstacles, and real-time sensed, possibly dynamic ones as
well as sensor states and other parameters. Planning requires a holistic view of the
vessel, all its subsystems and the environment in which it operates to make informed
and complex decisions. Examples of Planning include solving “Traveling Salesman” type
problems to optimize a set of objectives under various constraints in mid-mission.
Levels of Supervision
Having described the spectrum of environments and levels of vessel autonomy it is now
useful to define yet a third axis of consideration for meeting the tolerable risk