Now autopilots. While our SAS (with trim) is usuay good for VFR, for IFR there are stricter requirements about how much deviation a SAS actuator runaway is allowed to cause and the amount of deviation that can occur when the pilot is distracted. In some cases IFR is allowed with just SAS but it might come with a dual pilot limitation (so one of you will notice the huge deviation before it gets you!). There may also be a second series actuator (usually in the pitch axis) which is there to cancel out a runaway in the other one (again the EC135 pitch damper achieves this).
But if you want to do proper IFR you need more. Typically the minimum entry point is a heading and altitude hold but that's getting ahead of ourselves.
What we want next is for the attitude to stay exactly where we left it while we rifle through the Jeppesen or calculate a hold. We basically want a lower workload. We want attitude hold (ATT).
This needs more sensors (gyros or AHRS for each axis) plus magnetometers usually. Now the system can sense the outside world. It now senses the attitude of the aircraft, it compares this to what you wanted, then I commands the SAS actuators we already have to "stabilise" the aircraft around the new datum. Then it repeats, what you see is the aircraft bank or pitch to achieve/restore the chosen datum within limits set for how much attitude can be applied to do this. So what about the controls?
Remember the saturated SAS from earlier? This issue is now (usually) automated. The autopilot makes inputs through the SAS, but the autopilot constantly assesses the saturation of those actuators and automatically moves the controls through the trim actuator I mention above to recebtre the SAS actuator so it can do its thing. This function is called Autotrim on some aircraft.
So the autopilot always flies the aircraft through the SAS actuators not the controls (myth dispelled). The trim actuators can of course move the controls through their full range. If this ran away it would be catastrophic if these actuators acted quickly. So they don't. But combine short throw fast acting series actuators with a long throw slow acting actuator and you have the best of both worlds. Of note the trim actuators are mounted in parallel to the control run and so are sometimes called parallel actuators or rotary trim actuators or similar. Where are axis doesn't have a trim actuator, the pilot is typically required to move the control periodically to recentre the series actuator.
Now there's one problem with the aircraft holding the attitude. What if you want to turn. Some aircraft absolutely don't recognise the pilots inputs: the series actuator does its best but saturates. The pilot continues to fight the controls. The autopilot commands the trim actuator to move the control to get the SAS actuator back in the middle of its range but you're holding the control. The spring in the trim actuator then gradually builds up a force. You then release the control and BANG all that pent up force is released...rock and roll! The A109E does this.
However better aircraft always recognise the pilot makes an input and suspends the Autotrim function and attitude hold in that axis - effectively that axis drops into SAS (this may be called SCAS). It then seamlessly resumes ATT on release of the control. Pressing force trim release does the same but also removes the artificartificial
the autopilot can build on this underlying attitude system by building progressively more advanced (higher order) modes which use the underlying ones. So for heading hold, the autopilot commands a smooth series of attitude datum changes to achieve a nice turn. The navigation mode in turn commands a series of heading changes to achieve a nav route.
So deck landings. You want a stable airborne platform as possible to reduce your workload. Even if the ship is moving you want to stay stable relative to the world. Therefore ATT would seem the logical choice PROVIDED the system recognises pilot input. In this case SAS may be warranted but probably working against the trim (IE Force trim release not held down).
However immediately after landing the autopilot, ATT, SCAS, SAS needs to come off (or be automatically inhibited by weight on wheels/skids) otherwise the sensors I mentioned will sense attitude or rate change and attempt to damp the entire ship. The ship will win and the rotor disc will keep moving to try to damp the ships motion. Hello dynamic rollover.
I hope that answers your question. Happy to relate it to a particular type or variant of needed.