Introduction - Autonomous and Autonomic Systems - page 15

Though science of opportunity is not necessarily restricted by communi-
cation delays, it often has required a human in the loop. This can present
a problem since the phenomenon may no longer be observable by the time
initial data are transmitted back to earth, a human analyzes it, and com-
mands are sent back to the spacecraft. Autonomy can be useful in this area
because it enables immediate action to be taken to observe the phenomenon.
Challenges in this area include devising the rules for determining whether the
new science should interrupt other on-going science. Many factors may be in-
volved, including the importance of the current observation, the time to react
to the new science (it could be over by the time the instrument reacts), the
spacecraft state, and H&S issues. The rules would have to be embedded in
an onboard expert system (or other “intelligent” software) that could be up-
dated as new rules are learned. An example of a mission that does autonomous
onboard target-of-opportunity response is the Swift mission. It has a survey
instrument that finds possible new gamma ray bursters, determines whether
an object has high priority, and has autonomous commanding that can slew
the spacecraft so the narrow-field-of-view instrument can observe it.
For spacecraft H&S, a large communications latency could mean that a
spacecraft could be in jeopardy unknown to human operators on the earth and
could be lost before any corrective commands could be received from earth.
As in the case of science of opportunity, many aspects need to be taken into
account, to be embedded in onboard software, and to be updateable as the
mission progresses.
1.3.3 Interaction of Spacecraft
Spacecraft that interact and coordinate with each other, whether formation
flying or performing the science itself, may also have to communicate with a
human operator on earth. If the spacecraft does not have onboard autonomy,
the human operator performs analysis and sends commands back to the space-
craft. For the case of formation flying, having the formation coordination done
autonomously or semiautonomously through inter-spacecraft communications
can save the lag time for downloading the appropriate data and waiting for a
human operator to analyze it and return control commands. In many cases,
inter-spacecraft coordination can also save spacecraft resources such as power
and communication. However, the more the spacecraft interact and coordinate
among themselves, the larger the onboard memory needed for the state space
for keeping track of the interactions.
Whether formation flying needs autonomy depends on how accurately
spacecraft separations and orientations need to be maintained, and what per-
turbing influences affect the formation. If the requirements are loose and the
perturbations (relative to the requirements) are small, the formation can prob-
ably be ground-managed, with control only applied sporadically. If the require-
ments are stringent and the perturbations (relative to the requirements) are