Low Power Triangulation Circuits

Consider that the speed of radio waves is 1 foot per nanosecond !!

Location by time of flight is impractical on such a small scale as yours.

However, as few as two video cameras could locate an object in plan position and height above floor ( ie 3D) in a small room.

... Phil

Radio won't do, certainly not if your radio knowledge is as basic as you say.

Phil's suggestion of cameras is good, if you can do something to distinguish one "transmitter" from the other optically (big painted numbers comes to mind).

This sort of thing is done sonically to locate _receivers_, but I don't know if it's been done the other way around (although it could be). Basically one makes an "ultrasound LORAN" system. I've heard that these are on the market, but I've never tried to figure out where.

If you are allowed to build the floor you can make a system for each transmitter to sense where it is electronically and then just transmit it's position.

Ok, So if I understand this right radio waves moves too fast to measure small distances.

So it could work forslower waves right ?

As for optical detection, I ll check on google

Thanks for the input

well, differential GPS gets some pretty good performance using radio, works best outdoors though.

GPS hardware is highly specialised, and only cheap because of mass production.

for indoors using ultrasonics with either mobile beacons and fixed microphones or the other way round will probably be easier than using GPS.

You don't need transmitters, you just need a bunch of video or still cameras, one centrally located on each wall minimum, or more if needed due to field of view etc. Then it becomes simply a software and calibration issue. Code for image detection in video and still images is readily available. And you can use colour filters too if that makes your job easier.

Unltrasonics is another way, but that can get messy.

Forget radio transmission, it's too fast, you won't get such a small resolution.

Is this a real practical problem? or some uni project?

Dave.

Here is a company which sells the things you require:

Alternatively, if you must build something, then you could put six loops of wire, each loop tracing the circumference of one square face of your 3 metre cube room. At any one time you could drive one pair of loops on opposite faces of the room, first with in-phase signals, then with out-of-phase signals, then drive the next set of two coils, and so on. You could drive the coils with an audio frequency tone, e.g. 10kHz, and each coil could be driven for a few milliseconds or a bit more.

For each moveable receiver, I suggest using three mutually orthogonal coils (i.e. at coils in three planes 90 degrees to each other), which could be wound around a cube a couple of inches across. I used polystyrene foam and coated the assembly with araldite epoxy to keep the coils in place. You then need to amplify the signal from the coils with some low noise amplifiers, I suggest OP27 or NE5532 etc. You could either digitise the output of the opamps directly, or put the signal from each coil through a phase sensitive detector that is synchronised to the transmitter signal, low pass filter the output of the phase sensitive detectors and then digitise that.

From the vector sum of the signals from the three receiving coils, you can determine the amplitude and direction of the magnetic field relative to the receiving coils. If you do this when the two coils on opposite faces of your cubic room are energised in phase with each other, then this will tell you roughly the orientation of the receiving assembly. If you then reverse the signal to the coil on one face of your room then the field will have a gradient across your room and you can determine the position in the axis orthogonal to the plane of the two transmitting coils that you were using. You can then repeat for the other two planes.

You could then look-up from the initial position estimate some coefficients to correct for the field not being quite at right angles to the transmitting coils in some locations near the walls. As long as each location in the room gives a unique position estimate, you should be able to calibrate out the field non-uniformities using some look-up function.

You would want to avoid big lumps of metal in the room as these would cause distortions to the position estimates.

Chris

@ Chris Jones: Thanks for the info but i did not mean to actually build the whole system from scratch either :), but electrofields (have no idea how to translate the term in my language and dictionaries ain't helping) and me is really not a love story. I tend to dislike physics which is mainly why I kinda suck at anything related to waves.

@Phil Allison: If I get my stuff to work I ll be sure to answer why I want to do this :)

@David L.Jones: It is not a uni project and since I am working in a bank there is not really any direct real link to it. But same with Phil Allison, I ll answer why I want to do this either when I get it working or once I give up on the hardware part of this project which is the main step for me. And I would do this as a hobby :)

Bah, stupid of me not saying what I want to make. Especially since giving my need and not the solution would probably get me better answers anyway.

So I want to place for example 2 sensors in said room and follow the motion of it. But I want it to be wireless (because lines are damn annoying). I want the system to be able to operate a number of wireless sensors. Also, I am not interested in the position per say, but the motion.

To go back to technical stuff: So after my first post I tried to look at other stuff than radio. Ended up on accelerometers and they seem like they can handle the job. I don't really care about the position so accelerometers can do the motion job just fine and if need be even relate the position though with slight errors (rounding issues) Yes I know, commercially available for 2000$ for 2 wireless sensors. But:

- Less fun to do :)

- More expensive than the DIY way

- Won't learn how to write drivers for PCs

- Won't learn about how it works generally

- Won't be as customizable as I might want it to be

And here is what I am now thinking of building: Get for example a set of Accelerometers on own boards. Place them on an object lets say a club for example. Connect both sensors to a 'controller' that also has wireless capabilities on its board. It would be something like: Sensor ---- wiring ---- Controller + Wireless ----- wiring ---- Sensor Then have a wireless base that receive the transmission and send it to the CPU through USB for example. Then I treat the data as I want it.

The issues I have seen so far:

Other question:

If you see any other issues, have ideas on how to better that kind of system. Know of already build circuits and such (with documentation (at least a minimum)) it would be great.

Best regards, Chryzo

Bah, stupid of me not saying what I want to make. Especially since

At least one f*****ad can see how dumb it is not to describe the application.

He is still a complete f*****ad.

Cos he has chosen NOT to describe what is going on in the room that makes the sensors move about nor the purpose of the whole thing.

..... Phil

kes

lol :)

I won't describe the application whatever happens. I expressed a need with a bit of technological requirements, should be enough to get me going in the right direction. Though, found some stuff on freescale but it does not do exactly what i want :(.

You do not need to know about the end application to help though :)

Anyway, to answer my self, a ZSTAR3 system from freescale seems cheep enough to get me going though it is full wireless which isn't as good

Anyway

Thanks for the help and info :)

You can set up ultrasonic receivers, and chirp from the moveable object every few ms. If you also flash some kind of infrared LED when you chirp, you can use that as a timing reference. The relation between the time of the infrared pulse, and the reception of the ultrasonic pulse indicates the distance.

If you have one receiver, you can determine that the transmitter is on the surface of a sphere. Two gives you a circle in space, and three gives you a point.

Getting this to work will probably be very difficult.

Regards, Bob Monsen