[extropy-chat] Insect AI (was Wetware vs. Hardware)

[Adrian Tymes]

--- Max M <maxm at mail.tele.dk> wrote:
> But the major problem is not really the hardware. Even if we had it
> today, we would not know what to do with it. As far as I know, not
> even an insect has been modelled in a usefull way.
> 
> Otherwise my house would be filled with robots with insect
> intelligence for cleaning, and my garden would be minded by another
> bunch of them.

Actually, insects are being modelled today, although mainly for use as
toys and experiments.  See, for example,
http://www.jcminventures.com/Cybugs/cybug_~2.htm

I wonder whether it could be done, to have insect-bots programmed to
wander one's lawn (using sensors on the feet: if they exit a
grass-covered area, turn around immediately) and cut anything that's
not too big and hard (like a sprinkler head, a tot's or pet's limb, or
a tree trunk - just to name hazards in my lawn) at a certain
(programmable) height.  Random walk for half an hour (maybe less: you
want a swarm of these, and big swarms can do a lawn in the time it
takes to cross it), then return to the central hive (which does
recharging, minor maintenance, and alerts the user if the swarm is
getting low and replacements need to be purchased).

Of course, it'd be one heck of an expensive lawnmower at first, but
there would be a definite appeal to a certain early adopter market.
The challenge, though, would be to keep costs down: even said early
adopter market has its limits, and to truly mainstream this, the hive
plus bots would probably have to cost no more than $100, with
replacement bots (costing much less than $100) needed no more than once
every several months.  That, of course, primarily touches on
manufacturing and similar sciences related to the chassis - but it does
mean there can't be that much cost recovery for programming the AI, and
that the AI has to be fairly good at getting back to the hive rather
than getting lost in the field.