Newsgroups: comp.robotics
Path: brunix!uunet!cs.utexas.edu!sun-barr!apple!constellation!uokmax.ecn.uoknor.edu!rwmurphr
From: rwmurphr@uokmax.ecn.uoknor.edu (Robert W Murphree)
Subject: Re: Biological Plausibility (was Re: 6 legged beast)
Message-ID: <1992Aug6.194634.17166@constellation.ecn.uoknor.edu>
Keywords: mammals, brains, robots, ganglia, twitch, overthruster
Sender: usenet@constellation.ecn.uoknor.edu (Nets)
Organization: Engineering Computer Network, University of Oklahoma, Norman, OK, USA
References: <1992Jul28.233010.11281@constellation.ecn.uoknor.edu> <51096@dime.cs.umass.edu> <1992Jul31.232606.5964@constellation.ecn.uoknor.edu> <51292@dime.cs.umass.edu>
Date: Thu, 6 Aug 1992 19:46:34 GMT
Lines: 84

I apologize for shooting my mouth off. Ever since undergrad 20 years
ago I have studiously avoided knowing anything about the CNS.
Being a vestibular tech doesn't count.


My original reference was a longer article in Science News within the
last year.  don't ask me to look it up.
A recent reference is:
IEEE transactions on Robotics and Automation
June 1992, p 293-303.
"Robustness of a distributed Neural network controller for locomotion
in a hexapod robot"
by Hillel J. Chiel, Randall D. Beer, Roger E. Quinn and Kenneth S.
Espenscied.  all at Case Western Reserve, Clevland Ohio.

Abstract:
	A distributed neural-network controller for locomotion,
based on insect neurobiology has been used to control a hexapod robot.
How robust is this controller? Disabling any single sensor, effector or
central component did not prevent the robot from walking.  Further,
statically stable gaits could be extablished using either sensor input
or central connections.  Thus, a complex interplay between cental 
neural elements and sensor inputs is responsible for the robustness
of the controller and its ability to generate a continuous range of gaits.
This results suggest that biologically inspired neural-network controllers
may be a robust method for robotic control.

references in the paper cite other insect inspired controllers.


connolly@rabbit.cs.umass.edu (Christopher Ian Connolly) writes:

>In article <1992Jul31.232606.5964@constellation.ecn.uoknor.edu>,
>rwmurphr@uokmax.ecn.uoknor.edu (Robert W Murphree) writes:
> >But my opinion is they don't really understand the brain, bugs or humans
> >very well anyway.

>But this begs the question: Why, then, would one expect to find a
>"neurobiology of arthropods for roboticists" textbook?
how about a reader of old articles and excerpts that would be of
use to the mobile roboticist? How about bonehead neurology for 
electrical engineers. 
 
> >Bug behavior is simple enough that it seems to be
> >simulatable by modern control theory and computer hardware.

>Without a clear view of the neural substrates of insect behavior, what
>are you really simulating?  The most you can say is "See, my robot
>acts a bit like a cockroach." -- not a very impressive statement.  My
>opinion is that before one can say things like "robotic insect", one
>is obligated to provide some sort of clear model which explains *why*
>the thing should be regarded as insect-like.
The guy who told me you could simulate bugs behavior is dead, sorry.

> >The same cannot, in my opinion be said about anything vaguely
> >mammalian that has to do with the brain...

>Can you clarify this?  I can think of several researchers who have
>fairly specific things to say about mammalian behavior and its neural
>origins [1-3].  Alexander & co. at Emory have an interesting "multiple
>motor loops" picture of cortex, basal ganglia, and thalamus in mammals
>[4] that could provide fuel for a robotic analogue.
I bow to your greater knowledge. I did look at simulation of animal
behavior as a topic several years ago.  I think there are lots of
suceess stories of animal inspired circuits, neural nets to be sure.
I wonder if most of these aren't really more metaphorical uses of
someone's interpretation of how somethings' brain might work.
Some people try to simulate animal behavior like foraging, toads
mating call communication, I was interested in predator prey chases,
not too much has been done to really simulate somne species behaivor.
But it has been done.
It didn't look then (circa 1988)  that most simulations of animal
behavior were of particular use to controll theoristst

Of course neruscience gives us insight into animal behavior.  But do  
these insights amount to simulation?  Where you could take a film 
of the critter and compare it to your simulation?
Ron Brooks had some article I never read entitled "the whole iguana"
which talked about trying to simulate something really complex like
a lizard.  
>I don't dispute the notion that mammals are murky, but they're not so
>baffling as to warrant our complete neglect! (and I think this echoes
>the view that Ron Arkin posted here not too long ago.)
it doesn't pay to be too picky, does it.      
