Newsgroups: comp.robotics
Path: brunix!sgiblab!swrinde!cs.utexas.edu!uunet!news.ingr.com!ingr!b30news!ragtime.b30.ingr.com!les
From: les@ragtime.b30.ingr.com (les's desk)
Subject: Re: Future Robots in Industry (cont)
Message-ID: <1993Nov22.214436.5798@b30news.b30.ingr.com>
Sender: usenet@b30news.b30.ingr.com (Usenet Feed)
Organization: Intergraph Corp. Huntsville, Alabama
References: <752897110.F00002@ocitor.fidonet> <2cgj3cINN1cm@ymir.cs.umass.edu> <GAMIN.93Nov19125120@pellan.ireq-robot.hydro.qc.ca>
Date: Mon, 22 Nov 1993 21:44:36 GMT
Lines: 42

In article <GAMIN.93Nov19125120@pellan.ireq-robot.hydro.qc.ca>, gamin@ireq-robot.hydro.qc.ca (Martin Boyer) writes:
|> >>>>> 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.  
|> 
|> This is kind of short-sighted.
|> 
     Currently my mobile robot (OSCAR)uses "stereo" vidicon tubes with a pattern
overlay and can range focus from ~10" to 25' with an 1/16th at 42" which is
the range of the arm. The pattern is stored in a 128 bit number for repetitive
task. Built it over 10 years ago.
 
|> Specially in the field I'm working in, telerobotics, there are some
|> parameters that depend exclusively on the manipulator side of things.

	How would a hand capable of using the above vision recognize and pick
up a penny, a straight pin, a raw egg, or a 5 lb sack of sugar with no 
re-programming. The arm has a reach of 42" from the base. The main problem was
a combination of the vision and the tactical feel. The hand kept getting in the
way of the vision.

And there are no robots with the
|> versatility of the human arm.  
	OSCAR is capable of playing a polyphonic keyboard with three fingers
and a "thumb" I can still out play him but....
	
Robust software is still a research
|> issue.
	The concept of linear software is where most of problems seem to lay
with multiple processors working in a 3D parallel system normal linear compilers
just do not work very well. Each of OSCAR's processors (6502's) run a 2708 base 
"operating system" and a standard task packet off the central buses to perform
the task.

I agree that there is much research to be done in the manipulator area as well as
the visual area. The biggest problem that I see is the "thinking" languages that
are needed in the "what do I do next" from a reasoning perspective of the robot.
The need for language concepts that are non linear based are the "soft pieces" 
that need the most work.
I have rattled on enough you guys go back to your discussion.
