[ODE] OT: virtual creatures in games

[Nate W]

On Tue, 25 Mar 2003 david@csworkbench.com wrote:

> For example, if your creatures have the ability to move, then the an
> output of the genetic algorithm / neural network would be which
> direction and how fast, not necessarily what angle the knee joint
> needs to be, and the actual method of movement would be hard coded.

Hard coding movement is hard, though. :-)  That's basically what Juice is
all about, and I was kind of surprised how big a pain in the butt it was
to make something walk... and even the things that walk have a tendency to
fall over when faced with things like speed changes, or turning.

I'm pretty sure that a evolved (or "self-optimizing") control system is
the way to go for things that walk.  It could be useful to use hard coded
gaits to begin with, but I actually think that the amount of time spent
fine-tuning gaits would probably exceed the amount of time it would take
to write a self-optimizing control system.

If I didn't have a day job, Juice would probably be doing that sort of
thing by now.  If only it wasn't for that employer taking up all my
time... :-)

> Every time a new creature is spawned, create it with a new brain (I would
> suggest a genetic program over a neural network, as each node is more
> expressive, i.e. more intelligence for less calculation).  

There's a system called "fuzzy cognitive maps" that I've been looking
into.  It's sort of fuzzy-logic meets state-machine, and it seems like it
might be a good candidate for decision making for artifical creatures.  
Here's a link that gives a nice intro, and a relevant application of them:

http://www.ochoadeaspuru.com/fuzcogmap/WildlifePark.html

-- 

Nate Waddoups
Redmond WA USA
http://www.natew.com