NewScientist has an article about a British 007
spy snake robot can survive battlefield damage. The snakebot
is made up of many segments that move through the use of
nitinol muscle wire. If one of the snakebot's segments becomes
damaged or disabled, the other links can relearn its locomotion. The
snakebot has clevar built-in genetic algorithm software to rearrange its
undulating so that it can drag along a dead link, and keep on going,
albeit less gracefully and probably slower, to allow the robot to still
arrive at its
destination. The snakebot hasn't managed to slither out of the lab
yet, or really self-heal yet, or really evolve yet, but they claim that
Darwin would be proud of it if he were still alive. This
built-in adaptation concept should keep the robot at its
fittest as parts die off and prolong its survival until its
Nitinol Wire is very power expensive and low in pull strength and
I don't expect this robot will ever overcome those big obstacles
and become a real robot for use. If you look around to find out
which robots use nitinol wire, you'll find the "Stiquito" which is
a dinky robot that is tethered, meaning it doesn't have an onboard
computer or batteries, and is made
out of light foam core board, and still moves so slowly that
you fall asleep watching it. I think they've quit selling it
because it was so bad. Untethered robots that use
nitinol wire run out of juice quickly, go so slowly that you can't
stand it, and therefore don't get very far before running down.
I'm afraid this robot will be extinct in no time unless they
change the design and use some other mechanical means of locomotion.
Nitinol wire is not good for locomotion on a robot, but it may
be good for sensors or something where it doesn't require much power
or pull strength nor cycle time.
I bought a Staquito but never finished it. I got one leg finished just
to see how it worked and it was neat but too many requirements to work
properly for a larger robot.
It seemed that to make a real leg with any true use you really needed a
way to monitor the temperature of the wire to correctly decide how long to
apply current. It relied way to heavly on ambient temperature for me to
use it. If you applied current for too long you ended up with either a
smoking or in the least a damaged wire.
Too many requirements and not enough features in my opinion but still
worth playing with.
I've got one too, but I haven't cracked the seal yet. Haven't had the
time to play with it yet.
I was looking at the little Analog Controller they have, it is a
ULN2803 darlington driver with a 555 timer, that mounts on top of the
robot itself. Anyway, that controller can make it move forward at
different "breathtaking" speeds. (pun intended :) )
I lately have been thinking about using a Atmega 8 mini PCb with a
ULN2803 or 2003 driver to control it as well. Put a little IR object
detector on the front and it can move around and basically not bump
into things. Of course it still can't carry a battery pack, so it is
still tethered to an external battery pack. But then maybe it can
handle the weight of one 138mah 4v lion cell or maybe three or four
40mah nmh or nicad cells. I'll have to think about it some.