🚀 go-pugleaf

In article <1993Jan24.183226.10455@netcom.com> nagle@netcom.com (John Nagle) 
writes:

>       It's really time for somebody to start developing smarter sonar
> systems for robots.  We've been using those dumb Polaroid sensors for a
> decade now, and enough is known about bat, dophin, and submarine sonar
> to do a much better job.  The DSP chips are available and cheap.  
> Somebody with audio, DSP, or sonar experience needs to do this one.
> 
>       A reasonable way to start would be to use a machine with stereo
> audio inputs and outputs, and try to do bat-type FM sonar, operating
> around 10-20KHz.  
>
>					John Nagle
=================

About fifteen years ago I saw a movie made by --I believe-- Doctor
Leslie Keyes of the University of Canterbury in New Zealand.  It
showed blind children playing a block-stacking game while wearing
baseball caps fitted with Keye's phased-array sonar transducers above
the bills.  The transducers had about a dozen emitter/receivers
arranged horizontally to produce a narrow vertical beam that the kids
could sweep side-to-side by turning and tilting their heads to locate
objects precisely.  The kids listened to signals from the arrays using
stereo earphones (whether they heard the signals heterodyned down to
the audio range, or whether they heard beat frequencies from the
different transducers, I forget).  Anyway, the kids rapidly learned to
interpret the wideband, binaural acoustic signals to "see" the blocks
on the table in front of them, and were able to take turns adding a
block to the stack.  They could stack up about six blocks before the
stack fell over.  They placed the blocks on top of the stack
one-handed, without feeling their way up the stack.  This demonstrated
that they really were "seeing" the objects on the table with the sonar
arrays.  Other experiments showed that the kids could distinguish
different materials based on frequency-dependent scattering from
different surface textures, acoustic absorption characteristics, etc.
E.g., they could distinguish coarse from fine sandpaper, drapes from
walls, people from other objects.

Keyes had had a long and productive career developing secret military
underwater sonar systems, then decided to try to help blind children
with his skills after he retired.  In those days, the idea of even
approximating electronically the acoustic signal processing that the
kids learned to do with their ears was out of the question--and not
what Keyes was trying to do, anyway.  These days, inexpensive DSP
chips and neural network chips would make practical a great deal of
sophisticated processing of signals from such transducers.  There
would be plenty of applications for such technology in industrial
process control and inspection, robot guidance, and perhaps medical
applications as well.

I've mentioned Keyes' work repeatedly in different forums over the
years since then, but I've never heard of anyone's taking the
slightest interest in it.  If someone knows different, I'd like to
hear about it.  I think there's a lot of amazing technology like this
withering on the vine out there because the right person has simply
never heard about it.

Bill Park
=========

-- 
Grandpaw Bill's High Technology Consulting & Live Bait, Inc.