🚀 go-pugleaf

Many of those roboticists do juggle.  But the theoretical computer scientists
seem to be better jugglers.  I wonder why?  

Mohit asked, does one's own juggling transfer to one's robot?  Not clear.  But
studying biological systems can sometimes help guide design of robotic systems. 
In particular, Sakaguchi, Masutani, and Miyazaki (IROS 91) recorded data of two
guys, each doing two-ball fountain.  For each subject, the hand followed a
roughly elliptical path.  But the locations of the release and catch points along
the ellipse seemed to differ between the two subjects.  Sakaguchi et al designed
their robot to use elliptical paths too.

I've been juggling more intensely recently, partly in hopes of transferring
the skill to a robot.  It is too soon to tell whether I'm getting any
useful insights.  There are a couple of things that I could use some help on.

First, analysis and design of a catch would be a lot simpler if there were only
one strike, rather than a sequence of bounces.  One way to do that would be to
have a zero coefficient of restitution---a perfectly plastic impact.  Like when a
beanbag hits a concrete floor.  I've noticed that when I drop a lacrosse ball
into my hand, it doesn't bounce.  But if I repeat the experiment with my hand
resting on a table, the ball bounces.  Apparently flexion of the elbow joint
gives an effective restitution of zero.  But it only works with heavy enough
balls.  Nerf balls of the same size don't work.  It makes me wonder if very light
balls would make blindfold juggling impossible, or harder.

To apply the same trick to clubs, you need to cancel the translational and
rotational energy simultaneously.  In theory, this can be done by striking the
club at the center of percussion.  I notice that if I catch a club in my usual
fashion, but don't close the fingers, the club really does seem to strike without
bouncing, hang motionless for an instant, then slip off.  This suggests that
velocity-matching and closing of fingers play only a secondary role in catching. 
Any reaction?  It also suggests that clubs could be tuned "scientifically", by
matching mass and angular inertia to the dynamic characteristics of the human
arm.  Maybe even customized to the individual juggler, since some of us are
beefier than others.  Much as the "tuned track" was designed to match human
running dynamics.

The details of release are also interesting.  Baseball pitchers let the ball roll
off the end of the fingers to get speed and backspin, and they use knucklers to
get slow rotation.  But in juggling things are different because the speeds are
too low for aerodynamic effects.  (Aren't they?)  We care mostly about accuracy. 
How do the details of release affect accuracy?  Does it help to let the ball roll
out of the hand?  Do good jugglers all give similar rotation to the ball?