FPV is seen as an evil dependent link. If the pilot doesn't have his eyes on the drone, then somehow he'll be disoriented as to where it is relative to him. It may be assumed by the German government that drones under 5 Kg don't have sophisticated backups where ≥ 5 Kg perhaps do (or are mandated to have such??).
(In Canada (and the US?) FPV users must have an "observer" with them to keep the drone in sight at all times).
While "good airmanship" demands that we should be able to see, orient and control the drone (because it's in VLOS, right?) by sight alone there seems to be some notion that dependence on FPV will make for people flying far out of view and losing the ability to see the drone and therefore control the drone.
Why this makes sense: FPV depends on a data link which itself has various points of failure. If the pilot really can't navigate w/o FPV, then FPV (if it fails) means loss of control - how will the pilot find the drone never mind get it back?
Why this offends us: we (many of us) are secret RTH dependants. We are confidant that a properly configured RTH will get the drone back safely in case of RF loss or low battery or (forbidden thoughts ahead) pilot dependant RTH.
The main issue is that not all drones (and operators) are created equal. They don't all have RTH. They don't all have "good" RTH systems. Even when they do, people don't necessarily understand them or configure them properly.
And finally of course, it's dependant on GPS. Aye, there's the rub. Lose GPS while dependent on RTH and one may be done if operating on the fringes of VLOS.
/// beyond this point, I ramble ...///
Now DJI could include a complete dual redundant control system in their drones (they don't). But some drones such as the
Inspire 2 have 2 batteries and the system can fly with only 1. But it still has one GPS receiver.
Clever means of mitigation would be better integration of the (dual) IMU's with the (single) GPS and the (dual) mag compasses such that the IMU's alone could navigate accurately enough in case of GPS failure, aided or not by the mag compasses if available. All this takes is a few minutes of flying with a valid GPS signal. DJI don't do this - but it's very much in the realm of doable: the IMU's should be able to navigate back from a reasonable distance (1 - 2 km) with (much) less than 100 m error w/o mag compass and w/o GPS. But that's up to DJI to engage.
But even with all that nifty integration the government can never be sure that any given drone:
-is so equipped
-that the user knows how to set it up correctly
So governments reduce everyone to the smallest common denominator for the safety of everyone else.