Teaching robots to see

Intersting article at a site I read daily about an activity that most us assume is natural: seeing.  We assume that "seeing" is an in-born and natural ability, but it isn't: We have to learn to see, a process that dominates our infancy.

We look out into the world and see forms, movement, boundaries, color, light and darkness. We define the boundaries of a form by judging the gradations of light. We interpret motion by mentally sampling the perceived form against the background (which we completely understand to be a background), and we can identify that form specifically ("Bill") or generally (a man, but not Bill).

All this we do automatically, billions of mental transactions daily, and (usually) what we think we see is most likely what we think it is.

Robots, however, see our world as a two-dimensional canvas with a jumble of colors and shapes that bleed into a formless, amorphic blob.

I mention this because almost all near-death experiencers find themselves out-of-body, and to their astonishment, discover that they are able to see the same physical environment in roughly the same way as when they were in-bodied, despite not having eyes or even a brain.

Mainstream science rejects NDEs because scientists know perfectly well that this is impossible: our eyes see, and our brain perceives.

I think that this is a good point and worth wrapping our minds around it for a moment. Even if you believe that we "create our own reality" and that what we assume to be a physical is some sort of mental projection, why is the out-of-body reality perceived and recognized (at least in the initial stages) as being esstentially the "same" as it appeared moments earlier? Generally speaking, if the NDEr is in a car crash, that's what he sees
--the same objective car crash the his body was in moments earlier, in the same environment, with the same participants.

If, in fact, what we perceive as our physical reality is a objective representation, the same for all who see, regardless of how they perceive it, why don't robots see this automatically?