Just speculating; since this is not really my area of expertise, I'd guess that working from the FLCS maybe the route taken. Again guessing: would that not be the most economic way to work?
Personally...I would go for the stick and rudder pedals.
In a FBW-FLCS, the command-response loop is a closed loop. The cockpit create a command, it goes to the computer who does all the necessary computation from the command itself, accelerometers, gyros, and air data, sends that calculated command to the hydraulics, who complies and send back information on how much they displaced and at what rate they displaced. The FLCS computer then compares those responses against the cockpit command to see if the aircraft responded according to the pilot's wish, then the entire loop starts all over again when the FLCS computer recalculate any over/under response, which usually occurs but is compensated so quickly and fine grain that this continuous loop is never noticed by the pilot. This loop runs from engine start on the ground from take-off to engine stop on landing. If you ever notice the rear horizontal stabs moving up/down while the aircraft is taxiing you are looking at this loop. The uneven runway shakes the aircraft and creates gyros and accelerometers responses and the FLCS computer tries to compensate in leveling out the aircraft as if it is in actual flight.
Now...If I was to tap into the FLCS, I need to create two signals: A command signal and a nullifying signal.
The command signal would be to tell the FLCS computer to execute a maneuver. The nullifying signal would be to tell the computer
NOT to compare any responses against a supposedly command signal from the cockpit, of which I just simulated when I tapped into the FLCS computer.
On the other hand, if I tap into the control stick and rudder pedals, I need only one signal and that is to tell the transducers at the stick and rudder pedals that those controls have displaced by a pilot, which we know is not there. The rest of the chain as explained above will perform their roles without being the wiser that there is no one in the cockpit.
Do understand that what I explained is grossly simplified. I do not literally mean one signal but rather a set of signals because this is a quadruple (4) redundant system, from cockpit to hydraulics.
The test flight for the QF-16 had 7g maneuvers. Definitely exceeded expectations. Even more so that these Qs came from Davis-Monthan where they were pulled out of mothballed status and returned to flight. Seven times gravity of flight. Not likely any platform elsewhere in the world can meet, let alone exceed and this would be just another feather in the F-16's cap. So when they are fully certified to be training aids, they will be formidable ACM adversaries for our pilots. The F-16 have so many feathers of firsts and accomplishments that we might as well call it 'The Chief'.
