A Synthetic NERVOUS SYSTEM for Robotic Locomotion
Circuit 4. Variable master/slave central pattern generator.
lampreys — one might try opamp phase shifters (say 10 or
20°) to allow for smoother undulation.
Circuit 5 shows a simple amplitude modulator. This
allows one to control the amount of swing on the shaft of
the servo. Its input is OUT of the sine oscillator and outputs
to the DC modulator or directly to the input of the basic
motor neuron. This is useful in quadrupeds to allow them to
turn by adjusting the amount of swing in the front two legs.
With two light dependent resistors that are set up as voltage
dividers and a few resistors, a quadruped can have a light
following behavior simply by adjusting the front leg swing
proportional to light intensity.
Circuit 6 is a DC offset modulator that — when connected
to the output of the sine oscillator or its amplitude modulator
— allows adjustment
to the part of the
phase orbit that it
oscillates in. For
example, say your
sine oscillator is set
up for 2 volts, peek
to peek. With the DC
you can modulate it for a swing at 0-2 volts or 3-5 volts
(minus NPN voltage drop), which affects where the
servo shaft is swinging. (Is the control horn on the shaft
swinging on the right side or left side?)
This setup allows for positive input voltages for
both higher or lower offset, which simplifies circuit
design. This circuit is useful for tuning in two servo
walkers, swimmers, and crawlers and can be used to
adjust the balance point in a more complicated robot.
Circuit 7 shows a neat hack where you can replace
the AM modulator and the DC offset modulator with
the NE571 audio compandor — and get two channels, to
boot! The output of the sine oscillator feeds into pin 3
and a voltage on pin 1 controls the amplitude (not pin 2,
since this has a diode). Pins 6 and 7 are tied together and
are the output that feeds into the basic motor neuron.
Pin 5 with the resistor network controls the DC offset. The
other side of the chip is hooked up in an identical fashion, but
with a different sine oscillator feeding a different basic motor
neuron. Use the resistor network shown as a starting point for
inputs to the amp and DC modulators; add sensors later.
Pedal to the Metal
Circuit 5. Amplitude modulator.
Trace 1 shows the chaotic leg gait of a light following
robot that uses 34 NPN transistors and eight servos. The X
and Y motions of the leg were plotted on a storage scope
to show the up/down, left/right motion, as wells as
demonstrate the drift in amplitude over time.
The actual circuit, additional trace captures, and videos
showing this robot walking, are available on the SERVO
website ( www.servomagazine.com) for download. They
should give you a good idea of the power contained in this
approach to robot control.
Where to Go From Here
At some point, you might want to use microcontrollers for
Circuit 6. DC offset modulator.
Trace 1. Chaotic leg motion over time.
The Computation Brain
Chaos Theory Tamed
Locomotory Controllers in
Robots and Animals
(doctoral dissertation), Lewis
20 SERVO 07.2004