At least for known and somewhat controlled environments, tracking seems more or less solved, which applies to many industrial use cases. An example I've heard of: Lockheed and GE have been working on AR setups for airplane building and maintenance, with the idea that it becomes easier for workers to reference parts and look up steps to do, increasing training speed and reducing errors. Documentation says some part is broken, AR can tell you precisely where it is and potentially overlay diagrams with reality -> hopefully less chance for confusion between parts and less time spent looking up manuals. Maintenance worker checks a part and marks it as defective, the system knows which it is and there is less risk for it accidentally being written down wrong (personally, I suspect the current workflow is "look up/write in a paper manual", and you could get most of the way with a tablet computer, but why not go the entire way if tech allows it)

These examples work because a manufacturer already has a precise 3D model of the planes they make and thus can provide the data necessary for the tracking relative to the object, which is what matters in this case.

Apparently various beacon technologies allow sub-meter positioning in buildings/on equipped properties for relatively cheap, in cities GPS+WLAN can get at least rough locations. This could combine well with optical markers or simple image analysis for local, relative movements. That could be enough for navigation aids: an absolute error on a distance is not important as long as it properly tracks local movements. If it tries to guide you to a specific object you of course need better information once you get close.

Which again is interesting for all kinds of workers and potentially the general public. I would love a personalized guiding arrow when travelling through an unknown airport or other complex structure, or a rough target indicator in a city. I don't want to look down on a map on a phone, I don't need all the details, just something to keep track of. (until I think of the privacy implications probably...)

But I would not be surprised if the industry failed to make it simple, interoperable and sleek enough for public use anytime soon, so I'm not holding my breath for AR outside of specialist use and gaming, even though the basic tech is becoming viable right now.