Humans use stereo vision to overcome things like this. I think some kind of stereoscopic strategy would be a feasible strategy, no? Why revert to the age-old argument of "common sense"?
Humans have more than just stereo vision sensors. Like GP said, we also have robust schemata, seeded by biology and improved over our entire lives, that allow us to make sense of the things we see. These heuristics are trivial for us to access, but relatively difficult to describe and implement in a software system. That's what "common sense" means in this context.
I have one eye and have no problem not making these kinds of mistakes.
Stereo vision is overrated. One of my buddies is a pilot with 1 eye (since birth) and around 11 thousand landings. Top of his game. There is more to depth perception than stereo vision for sure
We have a lens that has to flex to bring far away objects into focus, so I figure that can give us some feedback.
For sufficiently distant objects, this is completely irrelevant. This lets us roughly gauge the reciprocal of distance.