Newsgroups: comp.robotics
Path: brunix!news.Brown.EDU!noc.near.net!black.clarku.edu!black.clarku.edu!kbasye
From: kbasye@black.clarku.edu (Ken Basye)
Subject: Re: Future Robots in Industry (cont)
Message-ID: <kbasye.753740630@black.clarku.edu>
Organization: Clark University (Worcester, MA)
References: <752897110.F00002@ocitor.fidonet> <2cgj3cINN1cm@ymir.cs.umass.edu> <CGqKrz.MGs@festival.ed.ac.uk>
Date: 19 Nov 93 20:23:50 GMT
Lines: 56

cam@castle.ed.ac.uk (Chris Malcolm) writes:

> buluswar@cs.umass.edu (Shashi Buluswar) writes:

>>I think one of the problems that is holding back the entire
>>field of robotics from an explosion into many areas of 
>>application is Vision.  

>If you keep very quiet, and you haven't been to many pop or rock
>concerts, you may just be able to hear a very high pitched burbling
>sound. It is the sound of a bat laughing.
>-- 

A couple of points:  

If the statement had been:

   I think one of the problems that is holding back the entire field
   of robotics from an explosion into many areas of application is
   imaging and image processing. 

I don't think the bats would have been laughing, or at least not any
louder than the dogs, cats, birds, mice and chimps.  We have a long
way to go regardless of what flavor the data stream comes in.  I'd
venture a guess that a lot of what we'll have to learn to do will
apply to many, if not most, kinds of images.

Second, I think the kind of explosion that Shashi is talking about
will require that robots be able to function in the kinds of
environments that *humans* find "friendly" (for lack of a better term)
and in which they have invested tremendous effort to construct.  That
includes the ability to read text, distinguish colors and myriad other
things that aren't possible without some "visual" input.  If having
useful robots around requires reshaping our environments to accomodate
them, there may be a small fire, but I doubt there will be any
explosion.  

Which is not to say that there isn't tremendous promise in the area of
non-visual sensing; there certainly is.  There is no reason why we
should not aspire to building robots with more and better sensing
modalities than our own.  But the fact is that our senses evolved in a
vision-dominated way for a reason, and our environments reflect that
dominance.  Sonar is great for deciding if that UFO up ahead is a
tasty mosquito, but not so great for deciding if that orange on the
tree is ripe.


kbasye@black.clarku.edu     
Ken Basye
Dept. of Math and CS
Clark University
950 Main St.
Worcester, MA  01610


    
