(wikimedia.org) EFIS instruments during a test flight.
How is the drift (accumulated error) nullified in INS-driven attitude/heading indicators?
The related post 'How are attitude indicators kept accurate?' does not cover the INS-driven flight instruments, whether electro-mechanical or pure-EFIS installations. In a purely mechanical system one of the solutions is the ingenious 'pendulous vanes' as the other post shows.
While there are two major types of INS systems, the strap-down and the stabilized platform, I don't know if the drift correction is the same (computational?) or different.
I'm asking out of curiosity and thus I'm looking for a simple (as opposed to inner workings) answer. For example, if computational, what inputs/sensors are used to correct for an unknown amount of drift? If mechanical, what is the equivalent to the 'pendulous vanes' when it comes to INS installations?
I've tried searching for an answer, but I keep arriving at very recent [and very technical] innovations/papers, and not the core principle.
Update:
I found this blog post, which is not about aircraft-grade INS systems, but if I understand it correctly, sensor fusion makes such instruments not prone to long-period drift. Is the same principle used in the INS-driven flight instruments?