Fully sighted, but I have been doing something vaguely similar for decades. As soon as I know an area decently well, or have taken at least one good look at a map, I can visualize the entire region’s roads and rotate it in my mind to find the best path (shortest time or least stress) from point A to point B. It’s fantastically handy to use in larger cities without having to pull out a map.
And as long as I have prominent landmarks to reference (mountains, etc.), I have a remarkably good sense of direction thanks to that mental model. Provided I have been fully aware of the scenery during my first trip anywhere, I can usually find my way back there from any other direction in which major roads exist. There is a fair bit of “hunting” to find the right road to go down, sure, but I rarely stray significantly from the correct direction.
Downside is that this is 100% reliant on sight to triangulate across geographical landmarks, I’ll grant you that. But it’s the mental model being used by the author that I’m surprised to see is so apparently similar.