Newsgroups: comp.robotics
Path: brunix!uunet!gumby!destroyer!ncar!netnews.whoi.edu!news
From: ulrich@canberra (Nathan Ulrich)
Subject: Re: CHALLENGING TOPIC ? IN ROBOT MECHANICS
Message-ID: <1992Oct21.224348.960@netnews.whoi.edu>
Sender: news@netnews.whoi.edu
Reply-To: ulrich@canberra.whoi.edu
Organization: Woods Hole Oceanographic Institution
References: <1992Oct20.122345.613@morgan.ucs.mun.ca>
Date: Wed, 21 Oct 92 22:43:48 GMT
Lines: 28

Raghu B writes
> Dear Netters,
> 
> 	I am just curious enough to know about the challenging
> topic in the Mechanics of Robotics that Major scientists or
> Engineers working on ?  I have always heard people saying
> that with proper control one need not even think of studying
> the mechanics of manipulators. How far is it True?

You can't make chicken soup from chicken s__t.

Anyone who thinks control, no matter how sophisticated, can fully compensate  
for inherent mechanical deficiencies doesn't know much about control.   
Literally hundreds of graduate students and academic researchers have tried to  
make Pumas (or other industrial robots) into proper force-controlled devices by  
the addition of a force sensor and "sophisticated control algorithms."  I have  
yet to see one succeed even marginally, yet robots like the WAM arm developed  
by Bill Townsend and Ken Salisbury at MIT can be easily force controlled at  
very high bandwidths....it was designed (mechanically) to be that way.

One way to improve the performance of a computer-controlled machine is to  
improve the machine.

--
Nathan Ulrich
Deep Submergence Laboratory
Woods Hole Oceanographic Institution
ulrich@canberra.whoi.edu
