Re: [mowbot] A heavy mower

Fred Thompson (fthompso nospam at
Wed, 16 Jul 1997 07:06:57 -0500 (CDT)

> >Could you please explain how I could determine direction with four
> >mikes in a square? I doubt I could measure any Doppler effects at
> >the 14 cm/sec current crawl speed.
> I just meant 'like a doppler direction finding array', not _actually_ a
> doppler array :-)
> By polling the 4 mikes and determining the signal strength from each
> you would be able to determine the direction of the 17khz signal. I
> realise that that is basically what you're doing now, but I got the
> impression that the current version wasn't as accurate as you'd like.
> Also using 4 mikes allows you to determine the robots orientation as well.
> Actually it is really a doppler system as the signal would be received at
> slightly different times due to orientation.

In Doppler DF radio finding, switching the antennas actually
induces a Doppler shift. When you shift your reception from one antenna to
one which is closer to the source, there is a phase shift which gives the
appearance of a rise in frequency. Switching to one further away applies
a phase shift which gives the appearance of lowering the frequency. In
doppler DF systems, you take 4 or more antennas and sequence through them
in a circular fashion, using an electronic antenna switcher (CD4066s).
The output of the switcher is run to an FM receiver. The output of the
receiver will have a "Hum" added to the normal signal received which
oscillates at the same frequency of your antenna rotation. A circuit which
watches this hum for the peak is also the one controlling your antenna
switcher. It turns on an LED for the antenna closest to the source.
You might be able to do the same thing in audio. I think what is
being described in the following section is more like a "Phased array"
receiver, rather than a doppler DF system. If your mics are placed far
enough apart you could compare sound arrival time to determine the mic
closest to the source. You could also compare the phase of the arriving
signals to narrow the angle to the source. There are some thorny issues
you will need to resolve to use this method. 1) the mics will need to be
matched so that the signal distortions introduced by them will be equal.
2) the mic dynamics will need to be modeled or understood when developing
your signal phase processing. 3) You will need some method to separate
the desired signal from the remaining noise.
(Note: audio filters introduce delays and phase distortions. For example
a tank circuit works by resonating at its tuned frequency. Every signal
component in the right frequency adds to its oscillation and out of phase
signals take away. It may take 10 oscillations of the input signal to get
the tank to resonate fully to the level of the input signal.)
Signal separation will probably require something like Digital
Signal Processing to get the desired bandwidth with consistan signal
delays. The problems with microphone dynamics might go away with a simple
training session (walk around the robot with a sound source in a wide open
field. Then create a look-up table associating computed direction with
true direction.
For more info on Doppler DF see "Transmitter Hunting, Radio
Direction Finding Simplified" by Joseph Moell K0OV, available from the
American Radio Relay League; 225 Main Street; Newington CT
06111;860-594-0200(, or your local library.

> Consider the following:
> 12
> 34
> S
> The signal S will be received first by mike 4 then 3, 2, 1.
> >>
> >> What are the chances of you mounting the XT on the robot to avoid the
> >> comms transmission problems?
> >>
> >Mounting a computer on-board is, at the moment, beyond my budget.
> >Good idea though.
> I am assuming that the XT is just for the robot. I'm sure you have another
> computer there. If you strip out the motherboard, IDE controller, Comms
> card, hard and floppy drives they could be mounted on the robot and a power
> supply constructed to run off batteries. I know of another roboticist who
> has done just this. I'll see if I can get some more information for you.
> Even if you don't want to do this I'm sure the information would be useful
> here.
> The other approach would be to use an old XT laptop, they usually go for
> bugger-all.
Read the adds in Nuts & Volts. It is not hard to find XT
motherboard for $15 - $20. I got one which includes monitor, keyboard,
serial, parallel and floppy controlled; for $20. With this board, you
could plug the computer into keyboard, monitor, and floppy drive. Boot up
and execute the robot program, then UN-plug the monitor, keyboard, and
floppy before sending it on its way. It will keep running as long as it's
powered. (My board came from: Electronics & computers Surplus City; 1490
W. Artesia Blvd.; Gardena, CA 90248; (800) 543-0540)

Fred Thompson
fthompso nospam at