> centimetre-level precision over four hours with a stationary setup

Ok, so does this really mean the smartwatch should not move for four hours?

Reading the paper[0], they're getting position data at 1s intervals, and ran it for 4h to show repeatable centimetre-level precision over that time period.

[0]: https://link.springer.com/content/pdf/10.1007/s10291-025-019... (note that the results using only the internal GNSS antennas are further into the paper)

Yes.

I mean it does demonstrate you can do it with something as small as a smartwatch