Alt-BEAM Archive
Message #05198
To: beam@sgiblab.sgi.com
From: Steven Bolt sbolt@xs4all.nl
Date: Fri, 9 Jul 1999 16:59:38 +0200 (CEST)
Subject: [alt-beam] Re: Chaotic controllers (was Something funny with the 1382)
On Fri, 9 Jul 1999, Bob Shannon wrote:
[ Emergent behaviours ]
> > > > The idea is to make such behaviours happen by design rather than
> > > > accident, but I haven't the foggiest how that might be achieved.
> > > > And everybody else appears to be just as much in the dark, so far.
> > >
> > > Actually the criteria for a system to exhibit emergent
> > > behaviors is fairly easy to define. If you study chaos
> > > theory, and chaos producing circuits you will find that as
> > > soon as we have 3 (or more) variables in a non-linear
> > > system, chaos is a given and emergent behaviors are produced.
> >
> > I'm aware of that, but afaik it so far doesn't translate to `bots
> > with increased performance, except by accident.
>
> I'm finding otherwise.
>
> The key may be that thinking of Nv's as nerons vastly
> overestimates the Nv. We should not be thinking of a photovore
> as a 2 neuron 'creature', its phototropisim is no better than
> many single celled creatures (that are phototropic in 3D).
No argument there...
> What I've done is to use the chaotic control systems in geometric
> arangements that match the physical body of the device. By
> selecting the chaotic attractor and geometery
> is fairly simple to get controlled behaviors.
>
> At least with the geometeries I've tested so far, which is very few.
Not intended as a `cheap shot', but that does sound a bit
Tildenesque. Nothing to do but wait for the details, and hope you
are on to something.
> Scalable in that higher ordered physical geometeries (of motive
> drivers) also extend the chaotic control circuitry without
> coherency problems seen in larger Nv systems.
Coherency is not the only and perhaps not the most important
problem of large Nv-like systems. Getting them to do something
useful, which can't be done cheaper and simpler in another way,
seems (to me) more difficult.
> But behavioral complexity is not a direct function of scaling the
> device and controller. Behavioral complexity is a function of
> the chaotic attractor and the geometery. There are a number of
> things in this design that are at odds with the thinking outlined
> in Living Machines.
Again, that statement doesn't mean all that much without the
details, but I hope it will deliver more interesting `bots.
Best,
Steve
----------------------------------------------------------------------
# sbolt@xs4all.nl # Steven Bolt # popular science monthly KIJK #
----------------------------------------------------------------------
------------------------------------------------------------------------
eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com
- Simplifying group communications
Home