measuring distance between two cars using infrared circuits

It might be easier with microwaves than infra-red:

Measure relative velocity using the Doppler effect. By integrating this, you get a running estimate of the change in distance. Weird things might happen when you go around corners!

Unfortunately, absolute measurement of short distances using electromagnetic waves is difficult / impossible due to the speed of light.

The SoL is a nanosecond per foot. A nanosecond isn't a particularly difficult thing to measure. Gate delays of modern semiconductor processes are in the few picosecond range.

No need for doppler measurements. Differentiate the distance calculations.

One can only imagine what you're trying to do. If you're gonna use the information to CONTROL the cars, you're absolutely (fill in the blank with your favorite expression of dementia). You're quite likely to get sued by the estate of the person(s) you kill using your contraption...assuming you survive. Suggest you get a different hobby. mike

One interesting, and perhaps simpler approach, is to take the same approach that the sharp IR distance measurement devices use.

These are cheap ($10) electronics gizmos that take 5V, and output an analog voltage that corresponds to the distance. They are not well suited to your application, because their maximum range is something like 80cm, but the scheme they use may be adaptable.

What they do is have an infrared LED, and a sensing device, which consists of an array of sensors. The LED and the sensors are separated by a fixed distance (maybe 3cm?), and arranged so that the returning IR falls onto a particular sensor according to the angle. Using this, they can sense the angle of return of the IR, and compute distance from that.

Seems like a similar scheme, albeit at a larger scale, might be usable.

Oddly enough, this is the scheme that bees' 'compound eyes' use. They don't have a 'continuous' set of receptors like we do focused by a lens. They sense changes with a set of cone shaped segments.

Feynman goes into detail about the compound eye in "The Feynman Lectures on Physics", volume I, ch 36. He makes some typically clever calculations, and determines that given a bee's eye, the maximal resolution vs the diffraction of light at the wavelength they care about will predict the shape of these cones. The formula he comes up with is

sigma = sqrt(lambda * r)

where sigma is the diameter of a segment at the tip, r is the length of the segment, and lambda is the wavelength of light to be seen.

This may actually affect the minimum resolution of your detector. Say your detector is w from the source, and you are projecting IR which has a wavelength of lambda. You presumably wish to detect differences at distances of about 10m. Say the resolution is deltaD. Then

deltaD = 0.5*w * (tan(t1) - tan(t2))

where t1 is the angle at the farther distance, and t2 is the angle at the closer distance. Then by a little trig, we have

deltaD = 0.5*w*(1 + tan(t1)*tan(t2)) * (tan (t1-t2))

since by Feynman's formula,

sigma = sqrt(lambda * r)

if the angle of the opening is (t1 - t2), and the internal length is r, we have

sigma = r * tan(t1 - t2)

so

r^2 * tan(t1-t2)^2 = lambda * r

and

tan(t1-t2) = sqrt(lambda/r)

thus, by the formula above,

deltaD = 0.5*w*(1 + tan(t1)*tan(t2)) * sqrt(lambda/r)

Now, assuming that w is 2m, and we want to detect differences at 10m, tan(t1) is about 10. Thus

deltaD = 101 * sqrt(lambda/r)

For lambda = 1um, and a detector of 2cm in length, that means the minimum distance resolution that can be detected is

deltaD = 101 * sqrt(1e-6/20e-3) = 750mm

before diffraction causes problems. This is 7% of the total distance! However, if you use ultraviolet, then you can make the opening much smaller, since the wavelength is smaller. For UV at lambda = 10e-7, you can detect

deltaD = 101 * sqrt(10e-7/20e-3) = 255mm

which is a bit better.

At smaller angles, the (1 + tan(t)^2) factor gets smaller. Thus, at 5m, we have 180mm, and at 2m, we have 35mm for IR with a 2cm r.

wrote

Usenet is not a write-only medium. You post your questions here and you get your answers here, otherwise the only person that gains anything from it is you. That's not the way a public forum works. If you'd rather work this issue via e-mail, then feel free to e-mail your question to as many people as you like.

Now to the question at hand. You need to specify the distance, weather, and other operating conditions that you expect this to work in. You are not likely to get this to work in the bright sun without allot of effort.

Or use something much slower than light - ultrasonics.

Well, radar was invented around 1939.

John

microwave is indeed more reliable using a bursting microwave gives an indication of absolute distance and speed between objects

timing between start of burst and start of reception of it is a measure for absolute distance doppler frequency gives relative speed

I can't help you with an electronic way of doing the task you require but I just happen to have worked with some computer software and hardware that can do this task very well.

What you are looking for is a "Stereo Vision Pair" (two cameras feeds) and some software that works out the distance between the cameras and a selected object. The software is a little complex but I have written some code that does the trick with simple math. This is all done using trigonometry. The most complex part of the software is object tracking, once you have zero'd onto the object you can take the displacement of the object from the two image frames and use trigonemetry to workout the distance between the cameras (which are normaly 90 degree and 6 degree with a spacing of 60mm) . It's very processor intensive but there are a number of algorithms that are required to do this task optimally. Once you have this all working you can even tell the size of objects and distance from with great accuracy (down to about 1mm depending on the total distance range required.).

I have a technique and system I developed from scratch, If you are interested I could give you a head start with information on object tracking, stereo vision metrics and optimizing algorithms for image processing. My system is developed using the C programming language and on the WIN32 API.

The only down side to this all is that the software is rather complex when you put all the components together.

Hope this may help.

for

schreef in bericht news: snipped-for-privacy@posting.google.com...

Mercedes-Benz uses a radar device in their adaptive cruise control system on their S class cars. I think it's not easy to build such device yourself from scratch.

Jeroen

I don't think that would work. Wouldn't there be too much other sounds, vibrations of air and weather to mess it up? Microwaves would probably be small enough and with a computer circuit you could take multiple readings per second and average it for more accuracy.

Actually I can just imagine that I wouldn't mind a device to tell me how far exactly the car is behind or in front of me. I wouldn't use it in repacementment of looking with my eyes though. But it might be nice for talking about those talegaters that come up behind you. Like "this guy actually got within 10 feet behind me on the highway" for example. Perhaps thats an extreme example though.

It is not difficult at all. It just requires bandwidth. Before anybody jumps on my case about detecting short CW pulses, let me point out that short CW pulses have a LOT of bandwidth, and the shorter they are the more bandwidth they have.

The same rules would apply to a modulated IR signal. There is no way the OP is going to get any kind of high resolution ranging using IR alone because there is just not enough bandwidth. (Some laser diodes have more than enough bandwidth to do this, but I don't think they put out enough power)

Ultrasound might work well. You could have a transponder on the back of the car in front and a range-finder on the front of the car in back.

--Mac

This won't work.

I suggest you try to think of a different approach. It seems to me as though radar is the best approach.

Ultrasound might work, but high frequency ultrasound attenuates rapidly in air.

You might be able to use two LED's on the rear of the car in front, and a video camera on the car in back. The distance would be calculated from the angular separation of the LED's. The LED's would have to be mounted with a carefully measured separation.

Good luck. You're going to need it.

--Mac

dementia).

Why? If better than a few inch resolution is needed, which doesn't seem to be the case here, use interferometry or as another poster said, stereoscopic vision. The guts of a few optical mice may do a decent stereoscope.

I'd think ultrasonics would be dicy in a noisy environment.

Ok, what's the bandwidth of a kHz modulated ~2GHz carrier (wherever there is some free bandwidth). It should be trivial to measure the round-trip delay to withing a nS, which is about six inches. At a kHz, that gives us a distance measuremnt every millisecond, which should be enough for distance and differentiate to give a relative velocity number.

Since it *is* done, I'm not sure why you contend that it can't.

It's *is* done without any transponder, which would make the idea useless.

Yes, pick your poision.

I was referring more to the RADAR range finders that Merc is using for "smart" speed control. IR diodes may have different problems (ambient noise, etc).

It also adds an unknown and significant delay into the path.