Newsgroups: comp.robotics
Path: brunix!news.Brown.EDU!qt.cs.utexas.edu!cs.utexas.edu!bcm!aio!news
From: WELLS <WELLS@CTSD2.JSC.NASA.GOV>
Subject: Re: End-point Sensors
Message-ID: <1993Feb11.225922.28916@aio.jsc.nasa.gov>
X-Xxdate: Thu, 11 Feb 93 22:59:56 GMT
Sender: news@aio.jsc.nasa.gov (USENET News System)
Organization: ER
X-Useragent: Nuntius v1.1.1d16
References: <1993Feb11.170114.2513@julian.uwo.ca>
Date: Thu, 11 Feb 1993 22:59:22 GMT
Lines: 44

In article <1993Feb11.193539.9000@sequent.com> Cory 
Carpenter, coryc@sequent.com writes:
>Couldn't you use one camera and some mirrors?  (See the 
crude ASCII
>graphics... .)
>
>               _____                    
>               |cam|=[ <<<<<<<<<<<<<<<<<<<<<<< Robot
>               |___|=[ <<<<<<<\                v
>                              ^\               v
>                              ^                v
>                              ^                v
>                             \^<<<<<<<<<<<<<<<< /          
>                              \                /
>
>Seems like you could split the camera's frame using the 
mirrors, so that
>you catch a direct view in half the video frame, and one 
from 90 degrees
>in the mirrors.

Use two mirrors, not three. The immage is reversed then 
righted with two, then reversed again with a third. There is 
also no rule of nature that constrains you to right angles so 
you can eliminate mirror 2 and still have a 90 deg view.

               _____                    
               |cam|=[ <<<<<<<<<<<<<<<<<<<<<<< Robot
               |___|=[ <<<<<* \                v
                                *              v
                                     *         v
                                         *     v
                                              * / 


You may also wish to consider one large mirror on the side, 
or even multiple mirrors on the sides of the robot target 
volume. This would provide some relief to the tight FOV 
created in the above two schemes. Again you have immage 
reversal, but this has been handled by scan reversal in older 
cameras or can be handled in the immage processing.

Consider that if you are trying to track the end point 
vibrations you're limited by the scan rate of cameras.
