tions and think of the nervous system as one big, black box.
This is actually a realistic approach, as we can walk and chew
gum while maintaining balance without conscious thought.
The brain doesn’t have to deal directly with some of these
lower functions, as can also be shown by the fact that a chicken
with its head cut off can still run around; the central pattern
generators for locomotion are actually located in the spine and
some reflexes are actually monosynaptic (knee jerk reflex).
Sensors interfaced to the control system can provide a
wide variety of adaptive behaviors, such as following light,
avoiding an obstacle, or shifting a balance point. Overlapping
of behaviors (subsumption) can allow for a wide range of
actions with minimal circuitry. Some sensors — such as light
dependent resistors — allow for a form of passive computing.
Once you have the circuits assembled to amplitude,
frequency, phase, and DC offset, modulate the signal that
feeds the “basic motor neuron.” You can think of this as a
neural cluster or a “synthetic ganglion” (ganglion being a
term from neural biology). When properly set up, it becomes
an analog computer that uses cheap hobby servos and offers
a large amount of processing power per transistor.
These are the basics of this nervous net idea. It should
be said that there is a patent pending on this concept. I’ve
done a lot more research that can’t be published at the time
of this writing due to patent reasons. Research is being done
Circuit 7. AM and DC offset modulator.
on self-organization, short and long term memory, hardware
genetics, and interfacing synthetic neurons to the nervous
system — all using commercial off-the-shelf-parts. SV
About the Author
Tom Jenner is self-taught in robotics and
electronics. His interests include artificial life as
it applies to robots, neural computational theory,
passive radar, and technical surveillance
counter measures. He can be reached at
Affordable Motion Control Products
Circle #135 on the Reader Service Card.
SERVO 07.2004 21