Reference: Immersive Design: The Next 10 Years of Interfaces
One thing I noticed from the reference article is the two limitations of Augmented Reality right now: Field of view and Input. Although the technology of recursive reality sounds fascinating, in most cases, such as the mobile App of AMNH Explore, our field of view is still constricted within the screen of a phone or a tablet. This constraint not only limits the user within extra flow such as downloading and installing applications, but also prevents them from sharing the experience with others.
From my own perspective, the most successful AR application which decrease this device constraint is InkHunter, an AR App letting users trying tattoos before they really go to the store. In this way, the phone screen is used as a scanning device, which decrease the gap between device and reality.
Speaking of input, the question would be "How do we interact with contents in a space?" Current solutions for this are still putting 2D or 2.5D interface into a virtual space. On the one hand, users won't feel strange when facing a traditional interface, on the other, the combination of 2D interfaces with a 3D space is kinda wired. Thus, I think the use of beacons would be a great solution. When a user interacts with AR contents in a space, they don't have VR glasses or handles, however, if we use position as an input to trigger new content, users may not have to press buttons or type in anything. In this way, we are able to decrease the distraction brought by input and make it a more immersive experience in the space.