twin vision

It is not clear what you intend to do.

But two cameras on one robot are normally configured to give a stereo image (with depth), to simulate the two eyes of a human.

Displaying stereo images is normally achieved with shutter glasses and with a software library like OpenGL. Use google for more info.

Bram Stolk

They do this all the time. One image is tinted in blue, and the other in red, and you look at it through glasses with one blue and one red lens. It does help if both cameras are looking at the same scene, of course ;-) - which is what I assume you meant.

Another thing is that you could probably create an image on a single screen that would look more 3-D than a single-camera image, by calculating parallax differences between the images from the 2 cameras, and playing some fun-n-games with shading or perspective, or something or other, during creation of the composite image. Not sure how well this would work, but it sounds something one could play with

- [besides using the alternating shutter approach]. OTOH, it might come out looking like something Picasso might have painted.

What I want to do is create the same sort of vision found in rodents, where the eyes are placed on either side of the head.

Gadget

Ahha. In that case, I think you should have a split-screen. How can you have 2 independent images without a split-screen? It would be a senseless jumble if you try to combine them.

Some animals [although I'm not sure that rodents qualify] have very little binocular overlap in their visual fields. Somewhere up high in the brain signals from the 2 sides will come together, but probably not in the visual system per se.

Predatory animals such as lions and eagles tend to have more binocular overlap, since it helps in running down prey, while prey animals such as deer and horses tend to have more laterally-oriented eyes, so they will have wider peripheral vision for better detection of predators. Rabbits apparently have almost 360 deg vision, IIRC. Dog eyes, evolved from wolves, tend to point more ahead.

Although rodent eyes may seem to point to the sides exclusively, I think there may be a small amount of binocular overlap directly dead ahead. And I'm fairly sure this is true in squirrels, at least, since they make their living jumping around in a highly 3-D world.

Rodents and similar "prey" animals do not have binocular vision -- they don't focus both eyes toward the same point. They have a very wide peripheral vision, almost like a Cinerama movie screen. The little rodent brain stitches the two images together, and it's basically a "split screen."

A robot could have two cameras, or an anamorphic cylindrical lens on a single camera. It doesn't matter to the image processor that the picture from the camera is all squeezed out and distorted, as long as the distortion is known. If you're needing the image to be presentable to humans, two cameras and a split screen would probably be best. Some simple video manipulation could "fuzz up" the split so it's not a sharp line.

-- Gordon Author: Constructing Robot Bases, Robot Builder's Sourcebook, Robot Builder's Bonanza