the hobby R/C PWM control pulse. Put 20 servos together
in one system and without a neural net that can learn to
adapt, all you can do is a slow static frame system.
Stiffness is one of the most important pieces of the
puzzle. If a joint is designed to rotate on one axis and
because of slop it moves slightly in another, then the
system becomes less deterministic. Slop is almost always
uncontrollable. It is unactuated and
very hard to predict. There may be a
time in the future when we use
unactuated movement to enhance a
complex system (like the flight of a
bat or an F-117) but for now, “if in
doubt … leave it out.”
This last rebuild removed
approximately 80% of the slop in the
We removed the pushrods
controlling the two axes of each foot
and replaced them with a simple
pulley and belt system.
We replaced the MXL80 belts in
the system with larger XL belts and
pulleys that are less likely to slip and
skip. We designed a new belt
tensioner system to allow easy
tightening of the belts.
We put all the pulleys on . 25 shafts. We now support
those shafts to prevent the very tight belts from deforming
All of the bushings have been replaced with . 25 ID
bearings to support heavy loads.
We also became more proficient with writing g-code to
run the CNC mill. Now every cut, hole, and shaping of the
SERVO 10.2013 57