[ODE] Re: Virtual creatures (karl sims etc...)

[Adam Rotaru]

--- Chris Campbell <chris.campbell@lincmedia.co.jp>
wrote:
> > With a little bit of cleverness, the virtual
> > creatures can extract quite a bit of energy from
> > this in a single time step.
> > For example, check out the video
> > CheatingJumper2_320.mpg here:
> > http://www.metahuman.org/martin/eden
> 
> hummm... ya reckon he did it on purpose? :-)


I think so.  I'm positive I can see the unmistakable
signs of malice and ill intention in the body
language of the creature.  BTW, I think it's a she.

More seriously, a very important theme is that there
is no purpose or goal in evolution.  No purpose
whatsoever.
This is a point which they try to explain to you,
you sort of accept it, but understand it for real 
only after spending a long time with your own
experiments.

This creature happened to jump up as a result of the
simulator inefficiency a bit more frequently.
Because the way the automatic selection was set up,
the cheating creature got duplicated.  Eventually
it improved, and in the end it passed the
experimenter's
conscious selection, and got rendered into a nice
movie
and posted on a mailing list.  Quite an achievement
for
a three-block creature.
One can say that the creature 'took advantage of a
particular
feature of the environment _in order to_ ensure its
survival',
but such a formulation can get misleading, as the
simple
creature doesn't have any sense of itself, or
objectives.

But before I digress to much, I have some comments
about
the specifics of the big jump which this creature
produces.
The effect of the contact forces (as described by
Martin) 
can be seen in the initial wiggling of the creature.
This is in itself a bit of cheating, as the creature
moves
around quite a bit.
However the big jump is different, and I suspect
different
reasons.  One guess is simultaeous interpenetration of
one body with more than one others.  This might cause
an unproportionally large aggregate contact forces.
Or there might be some other reason, such as
simulataneous
joint limit violation and collision (a wild guess).
I'm suggesting that the exact cause could be
determined,
and maybe mediated.  This is probably not an easy
task,
and there are many similar situations.

Finally, in passing I give a reference to an earlier
posting
of mine on another list, about requirements for a
physics 
engine for ALife experiments.  (This dates before I
knew about
ODE.)
http://lists.alife.org/pipermail/embody/2001q4/000025.html

cheers,
  Adam




__________________________________________________
Do You Yahoo!?
LAUNCH - Your Yahoo! Music Experience
http://launch.yahoo.com