Creating "local gps" system

Hello everybody,

I want to create "local gps" system.

The system will have 3-4-5-6 "satelites" (depending how much we will need, I guess it will be 3-4) that will transmit all the time unique transmit.

The system will have also "local gps" that will receive all the trasmits, calculate his position and trasnmit it to PC. When I am talking about "local" - I am talking about 3-10 meters area. Therefore the calculated location will have to be very accurate.

Is there any exist system like this ?
How difficult it will be to create this kind of system ? Any suggestions how ?
Does Accelerometer can supply this information ??

Thanks & regards,
Ofer

I think you're trying to make a triangulation system, which is pretty common. I'm not sure how accurate you need it to be, but you'll probably need to use RF or some of transmitter/reciever combo.

Read up on triangulation and/or GPS/GNSS.

I am talking about 3-10 meters area

I am thinking there is no way you will be able to do this. Work out the speed of light. 3 10-8 meters / second and you will see how accurate you will have to be. It's something that you can't just throw together. Most systems working over that range (like zigbee) use signal strength rather than flight time. The TI zigbee chip set has built in signal strength detectors to do just this but you need more than the minimum to do it with any reliability. The zigbee demos I have seen use at least 6 and possibly more.
If you want a ready made system look at the zigbee development kits, these usually come with 6 transmitters, but it is not cheap.

How about using ultrasonics? A ultrasonic speaker on one wall and another on the 90 degree wall, then by timing the pulses and how much they are offset from each other working out the position? (e.g. pulses set so that if you are directly in line with each speaker they are synced but once you move away they fall out of sync). Not sure if that is possible, just thinking out loud.

The problem with ultrasonics is that it will bounce of anything. So if this is for a little robot and you are walking round the room it will pick up you rather than the robot. Also I assume that it is the receiver that wants to know where it is, rather than the transmitter wanting to know where the receiver is.

Two sources aren't enough, at least for the general case. You'll need at least 3 to locate a point in 2D space, and 4 to locate a point in 3D space. More can be better, if your device can handle the math.

This is a seriously non-trivial problem. We had some students working on an AGV project that used 5 specialized radios that cost several hundred dollars each for a similar project (the space was ~30m square, not sure how far it would have scaled up).

-j

So I've been toying with this idea as well. I don't have hundreds of dollars for this (I'm thinking <$100). In any case, I was hoping to place several ultrasonic transmitters as the "satellites". Instead of having them on all the time, I'm planning on pinging them with an rf signal and timing the ultrasonic reply. The reply can be staggered (tx0 + 0ms, tx1 + 100ms, tx2 + 200ms, etc.) or encoded to distinguish them from one another (staggered might make the on-board processing easier, perhaps).

Now my problem is that I'm not sure where to get the transmitters that will give me sufficient range (5-10 m). Does anyone have any suggestions?

Another note: An estimate of heading may also be obtained by deploying two receivers sufficiently spaced apart on the robot/rover. (Again, this is a cheap alternative to gyros or compasses. I think Arduino can handle microsecond timing, so the short distance between the receivers - 10-20 cm - may not be a problem)

-Tom

Here are two other random methods to do a little thinking about: Wowee's rovio touts its "local gps" system as working very well. I believe all it is is a single light beacon from the base station that shines on the ceiling and rovio is able to triangulate itself in the room based on that.

Also, the blimpduino over at diydrones.com uses the pololu(sp?) modulated IR beacon for complete navigation. I think IR beacons are just about as cheap and low tech as is possible