🚀 go-pugleaf

     -=> Dodsonsl@memstvx1.memst. spoke of "Re: biological plausibili" <=-
    To All at 08-11-92  01:18
 Do> I intend to use analog processing, no cpu's, no code, just very simple
 Do> circuits that are based on Beer's model neurons that can be connected
 Do> together to control sensors and motors. This kind of architecture
 Do> should be quite scalable, perhaps more than Brooks' subsumption
 Do> architecure.

So are you going to use servo mechanisms?  [those would appear to
be the simplest way to use an analog output]

 Do> You (Stephen) also mentioned porting the code to a smaller machine. I 
 Do> think you should be warned that the code is 15000 C lines long. Each
 Do> model neuron is implemented as a differential equation solved using
 Do> Eulers' method. Each one (I used 85) is updated 4 times before the
 Do> physical body is updated. Of course, alot of the program deals with the
 Do> interface, graphics, etc. but it is still rather computationally
 Do> intensive.

hmmm so let us say that we have 85 equations each updated 4 x
before a movement.  The minimum "granularity" would be about 30
or 60 X a second so 1/60/85/4 =1/20400 meaning that each equation
must be solved within 49 micro seconds.  Or if you want to move
every 1/16 of a second 180 micro seconds.  For a 68hc11 @12mhz
would this be a real problem?  Can the program be "distributed"
among multiple processors?

        Stephen Cyberman@Toz.Buffalo.NY.US
        Scripted at Tue  08-11-1992  09:18:56

... Error 216: Tagline out of paper.
--- Blue Wave/QWK v2.10