I just tore down a broken DVD player I found on the street. It has three motors in it, two of which are geared to some kind of linear-motion slide; but all three of them seem to be just DC motors with no builtin servo.
Accurately positioning a linear-motion slide is a big part of what you need for various kinds of machinery, so I immediately got to thinking about how you could do it.
Aside from the most obvious things (the optical tape sensor from a printer, using a webcam, using the inverse square law with LED and photodiode) it occurred to me that the plastic slide itself probably has a relative permittivity of two or three, and so if you could arrange metal plates fixed in place on each side of it, you'd have a variable capacitor. Better, you could put one metal plate on its surface and another one fixed in place above or below it, with a layer of plastic between them keeping them from touching as they slide against each other. This would give you a capacitance varying from the femtofarad range to the 1000-picofarad range with the motion of the slide.
Charge-transfer sensing supposedly allows inexpensive measurement of variable capacitances to precisions of fractions of a femtofarad. If that's really true in this case, it should allow the inexpensive measurement of such a variable capacitor to some six significant figures, or 20 bits, of precision. If this were to be fully transferred to the positional precision of the slide — that is, if nothing else affected the capacitance than the position of the slide — then we would be capable of measuring the 13-centimeter motion of the slide to within 130 nanometers, about 1300 atoms. Sufficiently intelligent motor control, then, might be capable of positioning the slide to within 130 nanometers, although due to the many sources of slack and vibration in the system, surely not arbitrary motion paths with such precision.