ITAR has severely restricted access to nav system tech export, but there was always this open question about reckoning position by visual cues rather than GPS + dead reckoning. We all knew it was technically feasible, but we assumed that having “visual” cues was further out because AI at the time wasn’t up to the task.

There’s always going back to the old-fashioned method of having a trained pigeon inside of the device instead of a sophisticated computer vision system.