I have experimented with an optical mouse as absolute and relative position and
speed sensor. My concerns were drift and skew. Both drift and skew exist (plenty
of both, but low in magnitude and random) but for short term it is ok, when used
with an index sensor accurate to 1 mickey. Maintaining the target in the very
narrow focus depth of field of the stock sensor is very hard. Someone has
experimented with other objectives, with limited results (he used it for imaging
using the internal CMOS camera), the results are on the web.

The initial experimental object was a Maxwell pendulum used as target for the
mouse. I also tried a gravity pendulum and a long ruler. Short term throw and
indexing errors of 1% are common, and long term can be over 10%. There is also
"spill-over" into the other axis, I assume from orthogonality errors in the
alignment. The pendulums have the advantage of exactly computable speeds and
accelerations (from the period) and they can be tuned by adjusting the weight or
the torsion wire (monofilament fishing line) tension.

A more interesting application is an object tracker. Barring perspective change
caused image errors, an optical mouse sensor should be able to lock on to a
largish (based on its field of view) object and provide feedback steering
signals at a decent rate for a reasonably fast-moving object.

I used a USB and PS/2 capable mouse for my experiments, with an eye on
interfacing it to a microprocessor based project using the PS/2 mode. The
experiments were made more than 4 years ago.

  Peter


-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist