Alt-BEAM Archive
Message #06810
To: beam@sgiblab.sgi.com
From: Jean auBois aubois@trail.com
Date: Sun, 17 Oct 1999 03:23:37 -0600
Subject: [alt-beam] Re: Radio Communications?
At 07:51 PM 10/16/99 -0700, Senior wrote:
>Richard Caudle wrote:
> >
> > I've always thought that sound would be the optimum thing to use for
> > communicating.
>
>Oh true indeed. Check out steven bolt's work at
>http://www.xs4all.nl/~sbolt/
>and look at his spider's sound system.
Speaking of which, consider his reasons for using sound:
>For a small robot like the Spider, recognizing a great variety of sounds
>is out of the question. But it is possible to let simple hardware
>concentrate on a specific kind of `chirp', somewhat like a cricket. A
>chirping voice and well-tuned hearing would allow Spiders to tell each
>other what is going on, to get a good idea of the distance between them,
>and to track and find one another when they want. Worth looking into,
>don't you think?
I reckon that some of this is the whole point behind adding radio or sound
or whatever to a robot, whether it be BEAM or otherwise. In other words,
it is fine to speculate about how neat radio might be but unless there is
SOME expectation behind doing all of the work to implement it then it
appears to me to be mostly a lot of ... um ... speculation.
Having thus been a wet blanket, I'd like to turn around and have you
consider what the basic subject is -- COMMUNICATION -- and what purpose
does it serve, from the robot's point of view. The particular example I'd
use would be something nice and simple... say a mother cat and her kittens.
Generally speaking, the kittens stay around the mother (no doubt as a
result of imprinting which is an entirely different subject) but on
occasion one of them is certainly likely to wander off. Likewise, the
mother usually does a pretty good job of keeping track of her litter (even
to the point of knowing "how many" there are) but there are times when she
has to attend to her own needs like sleeping or eating and at those times
her attention isn't quite as focused. The net result is that there is a
kitten whose ONLY source of food (and no doubt protection, etc.) is
'somewhere else'. However, as long as the kitten isn't hungry, it usually
is pretty quiet.
Eventually the kitten becomes hungry; the mother knows one is
missing. From what I've observed, cats aren't particularly good at
tracking scents (something I found surprising -- dogs are hugely superior
in this ability but then look at the difference in the size of their noses)
... cats depend a whole lot more on vision and hearing. Well, the missing
kitten isn't visible, but because it is hungry it starts doing that little
"mieux" sound (it sure ain't the adult cat noise which is pretty definitely
me-ow, but that is another subject) at fairly regular intervals.
Now, this is the bit that I find the most fascinating. This is where the
communication comes in -- when the mother cat hears the kitten she responds
with her own noise. In fact, when closely observed, a dialogue (of a sort)
ensues: the two animals alternate. Now, obviously, the cats use the sounds
for binaural location, i.e. that is how they determine where they are going
next, but there is more to it than that. If the mother DOESN'T reply, the
kitten doesn't immediately call for her. It takes the kitten some time to
make a second call & at that point the subsequent calls become more
frequent. If everything works out right, mom finds the kit and takes it
back to the fold.
So... there has been a continuing discussion about the possible ecology of
BEAM robots which unfortunately often devolves into the "plants and
grazers" thread or the "gotta have some kind of nasty defense"
thread. Generally speaking, the "plants" thread involves unmoving light
gatherers and "grazers" that move away from the plant when they've got
enough energy and back (which of course requires some kind of direction
finding system) when they are feeling hungry. The "nasty defense" thread
is something that I believe is related to the age-related mindset of the
bulk of the BEAMers -- see Ian's survey to see where the peak of that
population is -- and I don't think it is useful for me to go into it.
But, why not a third possibility, one where there isn't a plant but rather
a 'mother' that does a really good job of converting energy (say, it has a
'huge' solar cell surface area) but it moves pretty slowly. Also, it is
only moderately good at finding where the best light pools are. Along with
that there is a 'litter' of 'kittens' (rather than grazers) that wander
mostly around the 'mother' (IR location == imprinting?) but occasionally
wander off. These 'kittens' would be pretty fast (unless they are nearly
out of energy) and might be useful for bringing information back to the
mother where a better light pool is -- a sort of scout.
In any case, our happy little family of robots succeeds best when it hangs
together more-or-less in a group. Unfortunately, vision doesn't always
work all the time -- one of them might find itself behind a chair or
something. As a result, sound is the backup system that provides the
necessary communication that supports success.
A lovely fantasy that... and Bolt's spider 'ears' certainly provide
something that might work as the front end. There is absolutely nothing
wrong with the back end of his system -- a microcontroller -- but it would
be interesting to come up with an analog (more BEAMish?) technique for
determining direction. Finally, once you've got all of these robots
chirping and listening and all that jazz, how do you bring THAT together
with the locomotion/behaviors required for success?
Just some late night blitherings from...
JaB
http://www.serve.com/heretics
An unconventional approach to the BEAM philosophy
Home