Searching \ for '[EE]: Determining vehicle speed based on sound' in subject line. ()
Help us get a faster server
FAQ page: techref.massmind.org/techref/io/audio.htm?key=sound
Search entire site for: 'Determining vehicle speed based on sound'.

Exact match. Not showing close matches.
'[EE]: Determining vehicle speed based on sound'
2001\02\11@214803 by

Hi all,

I'm doing a project in a DSP class to determine the speed of a vehicle
(like a car) based on just the sound which a stationary observer would
hear as it passes. I have a vague recollection that something similar was
mentioned here, but I can't find it in the archives.

Most importantly, I'm looking for recordings of various vehicles which
are going at a constant speed while passing, and (preferably) at a known
speed. So far, I have only found a few sounds of racecars passing, etc.,
but very few are going at a constant speed (I can hear them accelerating
and changing gears).

Thanks,

Sean

--

At 09:48 PM 2/11/01 -0500, Sean Breheny wrote:
>Hi all,
>
>I'm doing a project in a DSP class to determine the speed of a vehicle
>(like a car) based on just the sound which a stationary observer would
>hear as it passes. I have a vague recollection that something similar was
>mentioned here, but I can't find it in the archives.

ELTs are found by SARSATS that way. Started as a ham project IIRC..
The slope of the change in frequency as the sat passes, tells how far off
the sat is from a direct overhead pass.

The finding the zero crossing of that slope would let you take the
magnitude of the slope as the speed.

As a bonus, you should be able to tell which lane the car was in :)

--
Dave's Engineering Page: http://www.dvanhorn.org
Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9

--

Hi Dave,

I didn't know about the ELTs. I already have my method working for
monotonic signals (i.e., if a sine wave generator goes by, it can
determine the speed easily),the problem is that I don't have many
suitable recordings of actual cars (which is a much more difficult case,
since I have to determine speed from the engine noise and sound of the
air going over the car). I don't have any recordings where I know (from
some independent source) what speed the car was going.

Yeah, I could go out and record by the side of a highway, but I still
wouldn't know exactly how fast they were going, and I don't have enough
time before the due date to get a radar gun or something like that, or to
recruit other people to record a car at known speeds.
(Come to think of it, it would be gratifying to see all those brake
lights come on when I keyed the radar transmitter ;-)

Any idea where I can get such recordings?

Thanks,

Sean

On Sun, 11 Feb 2001, David VanHorn wrote:
{Quote hidden}

--

At 10:10 PM 2/11/01 -0500, you wrote:
>Hi Dave,
>
>I didn't know about the ELTs. I already have my method working for
>monotonic signals (i.e., if a sine wave generator goes by, it can
>determine the speed easily),the problem is that I don't have many
>suitable recordings of actual cars (which is a much more difficult case,
>since I have to determine speed from the engine noise and sound of the air
>going over the car). I don't have any recordings where I know (from some
>independent source) what speed the car was going.

Well, that's what I'm saying. It's not the absolute freq that matters, it's
the slope, and the amplitude of the slope.
You know the shape from your experiments.

>Yeah, I could go out and record by the side of a highway, but I still
>wouldn't know exactly how fast they were going, and I don't have enough
>time before the due date to get a radar gun or something like that, or to
>recruit other people to record a car at known speeds.
>(Come to think of it, it would be gratifying to see all those brake
>lights come on when I keyed the radar transmitter ;-)

Then they'd mung your data! :)

>Any idea where I can get such recordings?

EEk.. Hollywood, I'd guess, not cheap either.
Somewhere, there's a box of tapes with lables like "Car passing at 35, lane
2"

--
Dave's Engineering Page: http://www.dvanhorn.org
Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9

--

On Sun, 11 Feb 2001, David VanHorn wrote:

> Well, that's what I'm saying. It's not the absolute freq that matters, it's
> the slope, and the amplitude of the slope.
> You know the shape from your experiments.
>

Well, with the sine generator case, I can determine the speed without
knowing what freq the generator was set to. The problem I'm having with
the cars is that it is hard to determine what the original car engine
spectrum looks like. It isn't purely periodic, so there is some spreading
measurement. On top of that, most of the recordings I have are for cars
which were revving their engines up as they approach, and possibly
throttling back when they pass, so there are non-doppler induced frequency
changes.

Also, I'm not sure what you mean by "amplitude of the slope".

> Then they'd mung your data! :)

True, but I'd like to see that :-) I think some companies sell (used to
sell?) little 10 GHz CW transmitters in a small case with a transmit
button on them. You could go near a highway and just press the button and
see all the lights come on. Not sure if this cased any accidents, though.

> EEk.. Hollywood, I'd guess, not cheap either.
> Somewhere, there's a box of tapes with lables like "Car passing at 35, lane
> 2"

Yeah, that's exactly what I need. I used to think you could find almost
anything on the internet, but I'm not having luck with this. Any hints on
keywords to search?

Thanks,

Sean

{Quote hidden}

--

>
>
>Also, I'm not sure what you mean by "amplitude of the slope".

As they are coming toward you, you have F+doppler, and away, F-Doppler.
The difference is the amplitude I was talking about.

>Yeah, that's exactly what I need. I used to think you could find almost
>anything on the internet, but I'm not having luck with this. Any hints on
>keywords to search?

NO idea.. Very specialized.

--
Dave's Engineering Page: http://www.dvanhorn.org
Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9

--

On Sun, 11 Feb 2001, Sean Breheny wrote:

> Yeah, that's exactly what I need. I used to think you could find almost
> anything on the internet, but I'm not having luck with this. Any hints on
> keywords to search?

Usually anything except that *one thing* you really need... <grin>

Dale
---
The most exciting phrase to hear in science, the one that heralds new
discoveries, is not "Eureka!" (I found it!) but "That's funny ..."
-- Isaac Asimov

--

what are the possibilities of having a light beam sensor that recorded the point
at which the car passed the microphone. Then you would have a self calibration
point for each vehicle going past, and it would not matter if the vehicle was a
stock standard, been down the commercial tuning centre, or tuned by your local
backyard boy racer, each pass would have a self calibration of the "zero speed"
point.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

> I'm doing a project in a DSP class to determine the speed of a vehicle
> (like a car) based on just the sound which a stationary observer would
> hear as it passes.

Interesting.  I'd like to hear how well this works when you get results.  I
guess you will be looking for the doppler shift between approach and recede?

*****************************************************************
Olin Lathrop, embedded systems consultant in Devens Massachusetts
(978) 772-3129, olinembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

> Yeah, I could go out and record by the side of a highway, but I still
> wouldn't know exactly how fast they were going, and I don't have enough
> time before the due date to get a radar gun or something like that, or to
> recruit other people to record a car at known speeds.

Sounds like all you need is a camcorder.  It will capture the sound
directly, and it should be easy to determine speed by how far the car moved
over a few frames.

*****************************************************************
Olin Lathrop, embedded systems consultant in Devens Massachusetts
(978) 772-3129, olinembedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

Olin,

That's perfect! I just happen to have a camcorder with me (which is really
my parents' and would normally be with them) for reasons that have nothing
to do with this project! I'm charging up the battery now and I'm going to
try your idea. This is definitely one of those "why didn't I think of
that?" occasions.

Thanks,

Sean

At 08:30 AM 2/12/01 -0500, you wrote:
{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

Hi Sean,

I posted a link to a German site that had software for measuring model
aircraft speed using sound doppler (engine sound). Look up doppler in the
archive. Here is a copy from my archive:

Date: Fri, 12 Jan 2001 21:23:37 +0200 (IST)
From: Peter L. Peres <plpplp.plp.home.org>
To: pic microcontroller discussion list <PICLISTMITVMA.MIT.EDU>
Subject: [EE]: Physics: Doppler airspeed meter for models

Hi,

I just found this link, and I REALLY like it:

http://www.sprut.de/electronic/soft/scope.htm

It's in German. It describes speed measurements on aircraft models using
doppler (sound) recordings of a flyby played back to a computer program.

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

>The finding the zero crossing of that slope would let you take the
>magnitude of the slope as the speed.

Actually with sound you do not have access to the elements at 'infinity'
so the program does some pretty interesting asymptote calculations to find
those and then approximates the middle of them to find the pitch of the
engine, then returns to the asymptotes to find the speed from the presumed
doppler. It is not as easy as it sounds because the error is ^2 of
whatever error you measure the sound with in the first place afaik. I hope
that you plan to display the estimated error of the indicated value ;-).
There is also the effect of wind/air moved by the vehicle which can play
tricks with your calculations. I think that some experiments will be
required to compensate for this.

I don't know how it works for sattellites but the signals should be
significantly clearer and with both carrier and speed high, the doppler
signal will be significant and easy to use.

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

And if you plug the URL into Altavista's translator, you get the page
in English/Spanish, whatever.

http://world.altavista.com/

It's not perfect, but it sure helps get the gist of the page content.

And also look at
http://www.sprut.de/electronic/soft/vspec.htm
which does the doppler analysis in real time using the computer audio
card, rather than a .wav file as the below link does.

Robert

"Peter L. Peres" wrote:
{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

Thanks, Robert, for the links. You are right, you can often get the gist,
but I must say (from a great deal of previous experience with BabelFish)
that it is a pretty bad translator, especially with anything more complex
than a 1st year student of a language would come up with. For example, they
give links to newspapers which they say are usually translated well by
their translator. It is very funny to read the result of the translation.

Sean

At 04:58 PM 2/12/01 -0700, you wrote:
{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

You don't need tapes. Please read my other posting, take a tape recorder,
put it near the highway and drive past it at a few speeds (in both
directions). You can honk-code to mark the tape for passes and speeds.
Then go home and use the sound card on the computer etc. I think that you
can download the software from the German site and run a comparison using

hope this helps,

Peter

PS: How come time is so short for such a complex assignment ? This is not
something you get to do in a week ?

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics

I just had an idea. The problem of the changing pitch during measurement
can be solved if one uses a stereo recording of the event. By using two
microphones some distance apart one will obtain two sets of doppler 'null'
events and, more important imho, simultaneous recordings of the same
source with different doppler shifts. The extra data will supply the
information required to resolve the changing engine pitch and perhaps even
distance as David VanHorn said. It might even compensate for the car-moved
air. A stereo recorder is very easy to get and use and most laptops etc
also have stereo sound inputs. So everything is set imho. Working out the
equation set for this will be interesting imho.

Come to think of it, this can be used to clock car speeds using radio
frequency (one transmitter and two receivers on the side of the road, in a
line). No laser, no radar warning. Oops, speeding ticket. Especially since
the transmitter can be some distance away and be used for something else
normally (UHF TV station video H sync spectrum line comes to mind ... or
color burst spectrum line, gated).

In the next installment we will talk stealthed hotrod cars and doppler
jamming (just kidding).

Now, on the same theme, what happens if two cars approach each other while
passing the device and the wave bounces twice (tx->car1->car2->rx). I
think that some sort of vector addition will occur and the indicated speed
will be false and higher than expected ?

Peter

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservmitvma.mit.edu with SET PICList DIGEST in the body

>The extra data will supply the
>information required to resolve the changing engine pitch and perhaps even
>distance as David VanHorn said.

You only need one mic to get distance.

If the car is far, then the slope of the doppler near zero will be very soft.
If the car is close, then the slope is very sharp. More of a transition.

--
Dave's Engineering Page: http://www.dvanhorn.org
Where's dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listservmitvma.mit.edu with SET PICList DIGEST in the body

>You only need one mic to get distance.
>
>If the car is far, then the slope of the doppler near zero will be very
>soft. If the car is close, then the slope is very sharp. More of a
>transition.

You need the extra channel to supply pitch information with a different
doppler shift because the pitch of the sound changes for doppler-unrelated
reasons while it passes by (like accelerating/braking or simply change of
spectrum since you hear it from 'front' then from 'behind').

It is *not* as simple as with RF and it is not line-of-sight even at the
best of times. Hint: the size of the source is many times the size of the
probe. Even a 'sharp' zero is very wide and may be irregular. With models
you have problems if the engine is occulted by the wing when you measure
like this (f.ex. amphibian plane with pylon mounted engine).

Peter

--
http://www.piclist.com hint: To leave the PICList
piclist-unsubscribe-requestmitvma.mit.edu

More... (looser matching)
- Last day of these posts
- In 2001 , 2002 only
- Today
- New search...