[quotemsg=19118616,0,625763]It does need a really good algorithm to bend the image in the projection so that it can do the job. If implementing such algorithm would be so easy pokemon go would have had better augmentations too.[/quotemsg]First, there are surely limits to how much depth they can simulate, before you notice it's not really 3D.
Second, they use multiple cameras to scan the surface and estimate its depth. In this regard, you cannot compare it to Pokemon go, since the latter runs on mainstream phones which have only one back-facing camera.
Third, it uses face tracking, which I'll bet Pokemon Go is also lacking.
[quotemsg=19118616,0,625763]My thoughts where: Are you mostly just buying for the software, if it's made out of pico projector?[/quotemsg]It's not just a projector. It contains several cameras, which are needed to achieve the effect. This makes all the difference.
I'm thought these points were all pretty clear, from the article.
If this were inexpensive enough, I think it could find quite a few niches. I particularly like the AR for toys example, but both that and the educational applications will be severely hampered by its single-viewer restriction. Otherwise, if it could be made cheaply enough, perhaps some big toy company would buy them.