Class Project: Design a visual system for a robot headed to Mars
The problem of "designing" the visual system for a robot may seem daunting, vague and ill-defined - it seems that way because the problem is daunting, vague and ill-defined. But don't get overwhelmed or discouraged, for you'll begin to see the light as you tackle the problem. And don't misconstrue the purpose of the exercise. Our aim? To make explicit the goals of vision and the logic of the strategies by which those goals are implemented. Don't become focused on the particulars of the hardware for designing this visual system (although that hardware will eventually place limits on what's possible). Instead, concentrate at the outset on the nature of the problems that this visual system must solve. Carefully defining the problem sets the stage for planning the design. And once you do begin to think about design, take a very careful look at the design of one of the most effective visual systems in the history of life on earth: the human visual system. (Can we improve on what Nature has created?)
I like the following quote, taken from the influential book Vision (1982, Freeman Press) by the late David Marr. He characterized vision as "the process of discovering from images what is present in the world, and where it is." When thinking about how to tackle the problem of understanding vision, he wrote:
"There must exist ... a level of understanding at which the character of the information processing tasks carried out during perception are analysed and understood in a way that is independent of the particular mechnisms and structures that implement them in our heads.. Such analysis does not usurp an understanding at other levels - of neurons or of computer programs - but it is a necessary complement to them, since without it there can be no real understanding of the function of all those neurones." (Marr, 1982, Vision, Freeman Press)
We must first carefully define the nature of the problem to be solved by vision before we begin designing a visual system. Let that credo guide your initial thoughts about the project. And what better way to think about this than to ask, "How do I use vision on a day to day basis? What information is being registered by my eyes and brain, to allow me to accomplish those tasks?"