a optical mouse sensor could do head positioning with a stop sensor.

a optical mouse sensor could do head positioning with a stop sensor.
this would allow for belts driven by stepper motors.

What do you exactly mean? Throw the optical endstops away and use mouse sensors or use the mouse to count the teeth on the stepper motors?

The problems that will occur (that I know of) are the same as with the webcam:
1. Accuracy (totally depends on the CCD camera inside the mouse)
2. Programming
3. Skipping

The 1st problem can be solved, the 3rd one not actually... An optical mouse works by reading a 4x4 pixel field, and the software compares it with the next picture. If you have a blank page, the mouse sees a complete white space, even if you move the mouse. This will result in the mouse not changing directions, thus skipping.

If you want to put this one as a sensor for head positioning, then there are cheaper alternatives like an opto-endstop.

use a optical mouse and belt system and use the optical mouse to detect how far the head has moved compared to a paper strip colored so it is easy for the mouse to detect.
That would remove the need for lead screws and allow belts instead.

The Darwin already uses stepper driven timing belts. If you mean you want to eliminate the steppers in favor of DC motors using a mouse sensor for closed loop positioning, it's definitely possible - go ahead and give it a shot.

Currently RepRap uses belt drives (w/ opto endstops) for positioning in the x/y dimensions; lead screws are used for the vertical (z) dimension, but movement on that axis is slow and monodirectional and so there is no need for position tracking or corrections at this time.

Actually, I'm in the process of evaluating this idea. It will only work for linear sensing, not rotational sensing. My hope is that the skipped counts come more from the microcontroller attached to the sensor than from the sensor itself.

Optical sensors have come a ways since the ones that Mr. Seeker is describing. The unit I have is based on the ADNS-7050, which is a laser-based sensor. It captures a 22x22 pixel field and gives a resolution of 800dpi.

I'll let you all know how the test works out. My first attempt will be to prove out the repeatability of the sensor by measuring repeated moves on a ruler and establishing the error % for any given movement.

Brendan

I tip my hat to you I like your test procedure also be careful don;t lift the mouse too much or the lazor goes off... It happened when I wanted it to guide my robot.