Replies: 3 comments 2 replies
-
In terms of actions, some standard ones we could start with: measure this track; and search this area. I think the problem will also be working out how each sensor describes what actions it can do, in particular constraints, like what areas that it can search or measure tracks. For a stationary sensor these could be fixed, but for a mobile sensor, these may be time bounded. I think |
Beta Was this translation helpful? Give feedback.
-
Worth noting that SAPIENT-compliant sensors can report obscurations (one or more) as part of the heartbeat message. The ICD defines this as a region beyond a static obstacle so moving obstacle aren't supported, at least not yet. As per FOV, the obscurations are reported as range/bearing cones or location lists. I guess a good example of this would be if a vehicle parked up and partially obscured a sensor's FOV. After some time it can be classified as an obscuration. |
Beta Was this translation helpful? Give feedback.
-
For the operation of a sensor manager, we are thinking that
SensorManager.choose_actions()
should return a dictionary of the form{Sensor: [Action]}
whereSensor
is a sensor object and[Action]
a (list of) action(s). (This list would in most cases consist of a single action, but there may be instances where sensors can execute multiple actions per timestep.) This binds a particular sensor and action in the sensor management logic making it much more difficult to misassign actions when going through the combinatorics.The question/issue is, what form should
Action
take? It's likely something akin to resetting the internal state of a sensor immediately prior to making a measurement so should probably be undertaken by the sensor itself. Could we allowSensor = Sensor[Action]
, orSensor.take_action(Action)
, or something? (Sensor = Action(Sensor)
?)Beta Was this translation helpful? Give feedback.
All reactions