Footnotes

...
Visual space is most easily thought of as (but not limited to) a space of image (pixel) coordinates defining location and orientation of objects we manipulate, taken from one or more cameras watching the scene

...
Systems based on differential visual feedback are often advertised as ``uncalibrated'' or ``model free''. Of course they have some internal model while running. What is meant is that they somehow learn (or identifies) these models rather than requiring prior (analytic) models and exact calibration.

...
We use capital letters to indicate matrices, bold face for vectors, and plain for scalars and functions.

jag@cs.rochester.edu