I was thinking about the depth perception techniques we use for automated vehicles. The one that one the DARPA Grand Challenge used a laser rangefinder, as well as a camera; many of the contestants used lasers, radar, or a combination of the two.
It's generally well-known that human depth perception is provided by two eyes at an offset (binocular vision). There was one Grand Challenge car that used two cameras as its sole sensing equipment; I liked that solution, not only because of its simplicity, but because it modeled human vision. (They finished second, as I recall.)
However, wouldn't it be simpler to do it with a single camera?
Continue reading "Monocular Depth Perception" »


