Now a daft sounding question - If your plane uses a gyro based system for attitude indication - how does that system compensate for the curvature of the earth on long distance flights? You end up literally upside down when you land in Perth compared to when you took off from ORT.

As designer of many of these systems perhaps I can give a shortish description.

Start with the old vacuum driven artificial horizons. If you dissect them you will find a part called a "pendulous unit". This is really just a normal pendulum responding to gravity (or more accurately "acceleration"). If it is off kilter it opens vanes that force a change of airflow that spins the gyro so that over some time the gyro corrects.

This is also the reason why, if you would fly a continuous banked, balanced turn your horizon will eventually show that you are level even though clearly you are not.

Vacuum driven system where replaced initially with electrical gyros but a form of the pendulous unit would still be needed.

Modern AHRS systems started working in much the same way but relied on exceedingly accurate and drift free electronic gyroscopes to emulate what a good vacuum or electrical system could do and now used solid state accelerometer chips to emulate the pendulous unit.

Since electronic gyros that are good enough for this are still exceedingly expensive ways to use lower cost gyros have been created. Since these gyros tend to drift very fast on their own it has now become common practice to "aid". In principle we now use these gyros, accelerometers and something that tells us the velocity of the aircraft (true airspeed or a 3D velocity vector from a GPS or the usual sources). If we know the velocity and we know the rates of rotation around the three major axis (the gyros give us that with some random error) we can work out the acceleration forces acting on the aircraft caused by rotation and changes of the velocity. All the while our accelerometers give us the same information PLUS the gravity vector - we don't know by looking at the accelerometers where the direction of Earth is unless you are flying nicely straight and level. So now we can do a nifty little trick - we subtract the velocity derived acceleration force measurement from that of the real accelerometers and out pops the 1G vector that points to the Earth. It's "noisy" but the average is useful. We use that to correct our gyros continuously. This form of AHRS can show a correct bank angle regardless of how long you stay in a banked turn even if it is not balanced.

Now regardless of any of the above methods - we are correcting our gyro derived artificial horizon to detect and follow gravity. Since we are rotating at 15 degrees per hour (24*15=360) all we need to ensure is that we correct our gyros at a rate that exceeds this and we no longer have to worry too much about the Earth rotating.

Of course this only applies to those that are not members of the flat Earth society. For the esteemed members that do, no correction is ever required and any mention of such is at best a poor attempt of using observation to prove a fact.

Who said the sky is the limit ? I think not.