I’m, going to have to leave that for later in the week - I should have been asleep hours ago, and that looks like a looong bit of that most cumbersome-ly overcomplicating of tounges: Patent-ese… Hmm, I can’t even see the figures, just scrolling through. 
Is this patent for tech used in the Glyph, or more recent - like the alleged lightfield prototype?
I do notice that right at the top it says “curved mirror and partially transparent plate”, which reinforces previous impressions that the “retinal projection” label is nothing but so much marketing guff: So their projection screen is a mirror, instead of a matte or retroreflective surface - well yippety-doo-dah.
Not that I mean them and their stuff disrespect - just the up-propping, un-straightforward, purple prose (like in this very sentence :P). :7
If this solution allows you to focus naturally on different things, I highly doubt it’s entirely a software solution, because, you know, physics. -Besides: If it’s all software - what do they need the special glasses for?
If the software can render lightfields cheaply; That is absolutely super! Also absolutely necessary… But it can’t magically defy matters of real world optics.
What they could do with a custom combining lens, which you do say they have, could be to e.g. have it designed in such a segmented way that it takes partitions of the screen (just a grid of pictures taken from slightly offset positions), and projects them onto the mirror (visor) so that they overlap 1:1 on it, each emanating from its own position. -That would produce a lightfield; The picture coming from the emitting lens sub-part on the left will reflect off the mirror and travel on to the right, and vice versa. The question is a matter of the geometric solution: How large does the array of lens segments need to be to cover the required range of eventualities? (…which is not that broad, fortunately - the target that is our eyeballs is only so large, and we wouldn’t want to waste resolution on light that shoots off sideways where we’ll never see it, anyway. :D)
No, just that wide enough a range of them needs to be available at the same time, that our eyes have that whole range to freely choose from, without the hardware needing to monitor exactly what they are doing internally, or at least how they converge. :9
…and it is an extremely tight range, for our purposes - just enough to encompass the minute angular difference between light that comes at us from a fly that sits on a window a metre away, and the light that comes from the tree in the garden, right behind it.
That said: On a wider angle scale, you could have a single screen, with each half outputting imagery for both eyes, and achieve full human stereo overlap.
Welcome to the “CAVE” virtual reality system. 
There are of course multiple different ways to achieve for-all-purposes similar end goals - Oculus’ old.Half Dome mechanical varifocal research prototype, Magic Leap’s multifocal product, etc, etc. :7
Being able to display lightfields, either through techniques such as mentioned in this thread, or though holography, is one that is ridiculously density- and performance -heavy, but to my mind very elegant and powerful.
We play back Google’s lightfield captures in our headsets, but we do not see each frame as a lightfield – our view is of something that responds to HMD movement, but not to eyes accommodating or rotating.
You know what… I think the three of us are each arguing slightly different aspects of matters, and all of us erroneously conflating things the others say with things they didn’t necessarily actually touch upon. :7
.
Where do you live?