communication from UAV to pilot appeared relevant,
with pilots stating that it would probably be best if it
was situation dependent, but feedback would be
necessary in any case. Therefore, no clear
requirement can be formed here as opinions were
almost evenly split indicating the need to research this
question further.
Within the sub-category ‘Perceived advantages’,
it becomes apparent that pilots indeed see the
inclusions of UAVs as a positive development in the
field of military aviation as this could improve
combat power, lower the risk for pilots by reducing
the need for humans to enter hazardous territory, and
gather more information, thereby increasing SA.
Limitations, however, were also part of some
discussions. Here, pilots were mainly concerned with
the management of UAVs unnecessarily increasing
workload. One pilot stated:” We must avoid creating
more task load for the single remaining pilot up there.
The crew must make decisions and it cannot be that
the rest of the time, the pilot is busy making micro
adjustments”. Thus, the avoidance of micro-
management can be seen as a requirement of
collaborating with UAVs. On the other hand,
interviewed pilots insisted that critical decisions
should remain in their responsibility, as the creativity
in forming decisions is seen still as major advantage
humans have over machines.
Explicit requirements for this category are hard to
establish as opinions of pilots differ a lot, but what
can be derived is that pilots do not want to control
every little movement of the UAVs but instead take
on more of a management role.
4.4 Requirements for Adaptive
Automation
Within the category ‘Requirements for adaptive
automation functionality’, requirements for adaptive
automation behaviour were discussed in most detail
during the interviews. Here, especially transparency
was a requirement discussed in detail. Pilots agreed
that the actions of the adaptive automation have to be
transparent in order to trust it and not destroy their
SA. One pilot explained: “Theoretically, if the aircraft
told me that it changed from system state A to B, that
would be great, but if it wildly switches around and I
can not trust it, that would be bad”. Other aspects of
the transparency that were discussed were
predictability and reasons behind the adaption.
Another requirement discussed within this sub-
category is the ability of an override. Pilots
unanimously agreed that this is a function that has to
be present. Additionally, it was unanimously agreed
upon that the adaptive automation should not replace
the pilots but support them. In other words,
automation should not take away the control pilots
have, but take on more of an assistance function
supporting the pilot. To clarify, this does not imply
that the pilot cannot surrender tasks to the automation.
That is a prospective function very much appreciated
by pilots. Instead, the requirement deduced here is
that if the system does not keep the pilots in the loop,
it should at least support them in staying on the loop.
Furthermore, the topic of intelligent adaption
appeared to be relevant. For example, a rejection
strategy for multiple denied actions by the pilot was
mentioned.
The next sub-category (‘Trigger requirements’) in
line for being extensively discussed included
opinions of pilots in regard to which triggers to use
for the adaption. In relation to this, handling of
emergency situations was discussed in detail. The
handling should either include passively supporting
the pilot through the correct depiction of instructions
about what to do or actively intervening. Additionally,
pilots listed workload, flight phases, and external
factors (e.g. brightness of sun) as acceptable triggers
for adaption, whereas physiological parameters were
seen as critical by almost every pilot interviewed. “It
might be even more dangerous to monitor the pilot
and depending on this, change the HMI because that
will destroy his SA in a second”, elaborated one pilot.
Therefore, requirements for the sub-category ‘Trigger
requirements’ include adding emergency situations,
external environmental factors, workload, and flight
phases as trigger for adaption while excluding
physiological parameters.
A less detailed discussed sub-category was ‘Trust
& Acceptance’. Here, pilots either had the viewpoint
that reliability of a system function produces trust or
that a declining advection to the adaption (i.e.
auditive warnings first, then adaption proposals at
second, reduction over time to warning only) would
be helpful. However, others also stated that, since
their life is on the line, they are forced to trust the
system, therefore trust does not matter. The most
basic requirement unanimously agreed upon here was
that the system has to function as intended as,
apparently, the malfunction of systems was an often
occurring issue in past fighter models. Due to the
diversity in answers, no further requirement can be
formed here, therefore, further research into this topic
is recommended.
The last sub-category ‘Limitations’ included
drawbacks perceived by pilots towards adaptive
automation. Here, pilots appear to agree that adaptive
automation should not take away their responsibility