Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!rutgers!gatech!prism!jm59
From: jm59@prism.gatech.EDU (MILLS,JOHN M.)
Newsgroups: comp.robotics
Subject: Re: Getting narrow-beam range data
Message-ID: <31023@hydra.gatech.EDU>
Date: 10 Jun 91 13:45:44 GMT
References: <1290@sousa.ltn.dec.com>
Organization: Georgia Institute of Technology
Lines: 60

In article <1290@sousa.ltn.dec.com> smith@sndpit.enet.dec.com (Willie Smith) writes:
>
>In article <77904@brunix.UUCP>, kjb@cs.brown.edu (Ken Basye) writes...
>>We're looking for a way of getting a single range measurement in a
>>particular direction.  We have sonar, but the beam is far too wide.
>[...]
[objectives, questions, and suggestions deleted]
There are three ways to get better range:
  (1) more power (may fry transducer, but you could try to reduce the
      pulse or burst length, or otherwise limit duty cycle),
  (2) preprocessing or gain added to echo detector (but this is usually a
      "high tech" approach), or
  (3) narrow the beam, which it sounds like you also want for a more
      selective measurement.

Going with (3), two possibilities:
  (1) several transducers, spaced out into an array and driven in phase
      [another BUT] (but Polaroid may depend on the transducer's resonance
      to both form and resonate with the burst), or
  (2) use a parabolic reflector.

The other arguments were really "straw men."  I think a reflector is the
way to go.  For a spot beam, try a round reflector: maybe you can find 
one of those "spy microphone" toys once sold by Radio Shack and other
sub-MIL vendors, or the fire/cigarette lighter mirrors from Edmund, et al.
For a "fan" (narrow one way, wide the other), a long curved strip should
work.  Use parabolic shapes, and use a slide projector to illuminate
them for focus adjustment.  (a few "points of light" in the form of shiny
metal tape may help on a dull-surfaced reflector.)  You benefit from the
increased aperture on transmission and reception, which is important,
since you are fighting a 1/R^2 relation.  Once you have built the thing,
build a [fairly small] corner reflector, and map it out (unless you have
a microphone that can "see" the ultrasound).  Remember the fan beam will
be wide across the _narrow_ dimension of the reflector, and vice-versa.

The mirror needn't be much larger than the transducer's beamwidth, and
far enough away that the transducer and its supports don't block too much
of the view.  This sets the ratio of focal distance to reflector diameter.
(You will have to use a big enough reflector of long enough focal length.)
This implies you need a reasonable idea of the original beamwidth, but you
seem to have that.  If you want to get tricky, you can illuminate the
reflector obliquely.  You might get even get a product design out of this.

Consult books on antenna theory for more info.  Let me know how it works,
since these comments are in the "ideas are cheap" category: I haven't
actually built one!

>>Ideally, we'd like something that has low power consumption, a 5-10
>>degree cone (or less) and a range up to at least 4 meters, although it
>>need not be very accurate (even +/- 10 cm would be OK).  
>What's wrong with sonar again?  The Polaroid sensors have a beam width of 
>something like 12 degrees....

Regards -- happy hunting.  --jmm--

-- 
MILLS,JOHN M.
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp:	  ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!jm59
Internet: jm59@prism.gatech.edu