Micromouse?

[Chris Jerram]

==============================================================================
As a project I have to build one of the afore mentioned maze solving
devices. Has anyone on here done it before, and can recommend any
problems to avoid, give advice.

Thanks.
--
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Azerty Uiop. + EEE9...@IBM3090.BHAM.AC.UK +
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Education is a right, NOT a privilege. +
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[Thomas Ward]

Working on one at the moment ...

As a brief summary :
Most of the work seems to be in getting a good sensor system (and also a good drive
system). Many different CPU`s are used, with reasonable amounts of success for
most decent types (`Though there is an increasing trend towards higher-power
16/32 bit CPUs)

The best drive system seems to be GOOD DC motors with feedback (typically
optical incremental encoders with a few hundred pulses per rev)
The motors are usually driven with PWM to conserve power and minimize heat.

If you are just interested in getting a mouse working, then stepper motors
are a good choice - easy to control, but draw more power, less efficient and
harder to obtain.
The typical drive method is to have two wheels (one motor on each) and use
differential driving to perform turns, etc. This is a good method whether you
choose steppers or DC motors. There are also usually a number of castors (maybe one
at front and one at back) to give a stable design.

Sensors are typically `Over-the-wall' with IR emitter/detector pairs (maybe 6 to 15 pairs on each of
four arms - one to each side at both front and back). They look at the reflected IR from
the top of the walls, but be careful of varying sensitivity detectors causing eroneous results.
(some detector devices have a WIDE range of sensitivity variation between parts)
Ultrasonics have also been used with some success, but the general opinion is that it is hard
to decode reflections from various bits and pieces within the maze. (You need a LARGE sensor
to get good angular resolution) Some good work has been done with CCD devices allowing a
below-the-wall sensor system with better resolution, but this is generally harder to get working
than conventional approaches. - I`m working on this one at the moment.

Power source is almost always NiCd, allowing reasonable storage with good load-bearing
capability. Alkalines are possible, but don't allow as much current to be drawn. Lithium
sources have also been used (Mitee II/III etc.) but are expensive, and typically low-current as
well.

A really good article on MITEE MOUSE III has been written in `Circuit Cellar Ink' a few
years back - well worth reading.

Hope this has been of some use,

Tom Ward