Re: Isn't this inertial guidance?
Half right Charles.
Integration itself is error prone, we don't have infinite precision in math processing itself - we have floats, doubles...and on up to a point, but it's not enough as errors accumulate - it's integration. Done over small enough sample intervals to whatever precision the buildup of error will kill you with any known processing power that is or likely will become available in the next few hundred years. See N body gravitational problem which has to be solved by perturbation (there is no closed form feedforward solution) and the inability to predict where, say, Jupiter will be in a hundred years to better than a bunch of miles accuracy. And that's with known inputs to any precision you'd like and things that take years to move around in a well known elipse - or close, as they pull on things that move and pull back and so on.
Same class of problem here.
In this case, the initial measurement by accelerometers, even if utterly exact (infinite bits/sample!), will still give errors due to gravitational anomalies making a thing that measures acceleration think its drifting up or down depending on what the gravity is where it happens to be, versus where it was calibrated. For example only - the field can be skewed in other directions as well. Looking at that is is one way to find things underground as a fairly well developed technology all its own.
Slight changes in gravity even affect clocks at this precision level: https://physicsworld.com/a/a-brief-history-of-timekeeping/
Yeah, for that, with a 100% accurate gravitational map (NASA is making them) you could correct. Thing is, at this level of accuracy, the tides and so on become significat sources of error. This is just more press-release "gimme another grant" technology with a hint of some science attached.
Biting the hand that feeds IT
