My astro code works pretty much like Bill's, other than the square root
business, which could be spliced in by replacing:
ex = rx - px; /* position difference error */
with
ex = root(rx - px)
where root(x) is defined as
root(x) = sign(x)*sqrt(abs(x))
and sign(x) = 1 if x >= 0, -1 otherwise.
So its just a twist in the perceptual function that lets astro close
in faster on mother without overshooting. If there were
self-reproducing robots whizzing around in space, natural selection
could cause them to acquire this function, since its utility follows
from fixed properties of the environment.
I'm beginning to acquire some confidence that astro's reorganization
really does confine itself to a region for k2 (in Bill's C code),
fluctuating between .2 and .9, where about .55 seems to be about
optimal. I think the fluctations are due to the intervals between
choice of new delta being too short, so that differences between
current and old average error are too much influenced by accidents
of how mother is moving around. But I think I'll have to redo it
properly in C (it's currently Turbo Prolog (!!! ... because of the
nice IDE)) to get it to run at a reasonable rate.
From an empirical point of view, does anyone know anything about the
distance - velocity functions used by birds when landing? It might be
interesting to see how living systems actually tackled the problems posed
by inertia in a low-friction environment.
As for Alife, etc: Many of these systems (and also Chapman & Agre's
video-game playing programs) do model significant aspects of keeping oneself
alive (that's why video games are fun). So either they are in fact full
of control systems, perhaps to a greater extent than their creators
realize, or they are leaving out aspects of reality for which control
systems are essential. Either way they provide lots of stuff for people
to do, either in the way of improving our understanding of how they
work, or in making them more lifelike, or both. This is pretty much
what Greg Williams has been saying, I think, and of course his & Pat's NSCK
program is a fabulous resources for helping people to get on with it.
Avery.Andrews@anu.edu.au