Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

precise positioning system

Status
Not open for further replies.

mitgha

Newbie level 3
Newbie level 3
Joined
Aug 25, 2012
Messages
4
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,328
Please help me to have an idea if the following is possible:

I need a positioning system for flying small aircraft,
I want to position them with high accuracy and fast.

The specification would be as follows:

High speed: about 20 reads per second!
Hight precision: 1cm
Reliable
Outdoors but in open field, no buildings.
Area covered: 2km by 2km.
Total budget: 5000$

My first thoughts are about an RF system.
Transmissions done from the aircraft with couple of receivers
positioned on a grid of 500meter on the surface.

Thanks for your help.
 

I think in theory you can find the position with just 4 receivers. In practice, an accuracy of 1cm may be difficult. An accuracy of 1m should be relatively easy, though.

How small is the aircraft and what sort of altitude will it be flying at?
 

I've done some work in this field. You can use triangulation but it will be far easier to use GPS. The problem with triangulation is you need a minimum of three receivers and they have to be widely spaced apart, I'm guessing you want something that can be used from a single location. Also please explain the need for 1cm accuracy in 4kM area, is it just for controlled landings? If so, a diiferent method of assessing altitude should be used.

Brian.
 

Also please explain the need for 1cm accuracy in 4kM area, is it just for controlled landings? If so, a diiferent method of assessing altitude should be used.
Ultrasonic transducers may be good for that. I've seen them used in a toy helicopter for stupid-proofing.
 

The sound speed is slow for 1km scale, it takes about 2.9 seconds for sound to travel 1km in air!
The other left option is some sort of electromagnetic waves (Radio, microwave, infrared, visual light, uv, ....)
1cm is the wavelength of 30GHz and for a good filtering one needs even higher frequency I guess.
So we come to the range of 300GHz or so. This is the highest microwave frequency and the lowest infrared!
So an RF (Radio) system is also not possible. Microwave might be but I don't know about these much.
Another factor we should consider is the cost, and it would be good to use consumer parts so to bring down the costs.
After microwave we are should design a system with infrared or visible light.
I like visible light option. For example an HD camera with good lens which has very good zoom.
Then I mount these on a precision servo control. I will put some 2cm diameter red dots on the airplane to
follow and the camera using a software will follow them. Using the relative position of the points on the
picture I can get the position and direction. Then from change of position and direction the speed, angular speed,...
 

mitgha said:
High speed: about 20 reads per second!
Hight precision: 1cm

mitgha said:
Then I mount these on a precision servo control. I will put some 2cm diameter red dots on the airplane to
follow and the camera using a software...

I think you are proving it won't work without some change in specifications. 20 reads per second = 50 mS between reads. At 4 km/h (extremely slow for an outdoor airplane), it is traveling 111 cm/sec. That is, it will move 5.6 cm between reads. You will not be able to track it.

John
 

I hear some unvalidated assumption about a relation of resolution and frequency of RF measurement systems. You can easily refute it by looking at industrial measurement systems, e.g. FMCW radars.

An optical solution may be nevertheless reasonable because it doesn't depend on a precise timing reference of receivers.
 

why I can not track it?

https://www.youtube.com/watch?v=iRlWw8GD0xc

- - - Updated - - -

I hear some unvalidated assumption about a relation of resolution and frequency of RF measurement systems. You can easily refute it by looking at industrial measurement systems, e.g. FMCW radars.

An optical solution may be nevertheless reasonable because it doesn't depend on a precise timing reference of receivers.

Well you might be right.
But anyway for reliable 1cm resolution I think one needs about 30GHz at least.
 

That YouTube didn't give the frequency of sampling (it was silent on my PC). If your sampling frequency is less than the potential change in position, then the camera won't know were to move for the next sample. Your 2-cm spot will be gone within the sampling time. Airplanes do not follow simple ballistic paths that can be predicted. (Some of mine seem to, but that is a different story. ;) ) The camera will need to recapture. For an airplane moving a more realistic 30 km/sec, with average wind speeds of 10 knots or so, clouds (?), thermals, and other turbulence, the camera might spend most of its time searching, and not finding. If it is looking for a simple reflection, there are other things in the air that reflect besides the model you are trying to control. That could give you false captures.

Assume the camera can capture in 20 random samples (as you state in your specifications). That is one second. A lot can happen in an airplane in 1 second. Of course, the human eye, when controlling an airplane at a distance is capturing a much larger field than a 2-cm circle and is probably not controlling to a precision of within 1 cm.

I have made a distinction between outdoor models and indoor models. The latter can fly at quite slow speed -- slower than a slow walk.

John
 
That YouTube didn't give the frequency of sampling (it was silent on my PC). If your sampling frequency is less than the potential change in position, then the camera won't know were to move for the next sample. Your 2-cm spot will be gone within the sampling time. Airplanes do not follow simple ballistic paths that can be predicted. (Some of mine seem to, but that is a different story. ;) ) The camera will need to recapture. For an airplane moving a more realistic 30 km/sec, with average wind speeds of 10 knots or so, clouds (?), thermals, and other turbulence, the camera might spend most of its time searching, and not finding. If it is looking for a simple reflection, there are other things in the air that reflect besides the model you are trying to control. That could give you false captures.

Assume the camera can capture in 20 random samples (as you state in your specifications). That is one second. A lot can happen in an airplane in 1 second. Of course, the human eye, when controlling an airplane at a distance is capturing a much larger field than a 2-cm circle and is probably not controlling to a precision of within 1 cm.

I have made a distinction between outdoor models and indoor models. The latter can fly at quite slow speed -- slower than a slow walk.

John

Ok let us assume a high speed camera (well this is not really a good option because it is not a consumer product).

1216 x 1216 resolution at 1,000 frames per second!

Let the zoom be such that the width of picture corresponds to 8meters.

8meter/1216=6.57mm, so a red dot of 8cm diameter will correspond to 12 pixel in the image.

If the airplane flies 60km/h=16.6m/s. So between two frames the airplane will move 16.6mm
which corresponds to 2.53pixel.

So on the picture the software should follow a 12 pixel in diameter circle with 2.5 pixel movement per frame.
I think this is possible.
 

The specifications seem to be drifting. I thought you said a 2 cm spot and 20 samples per second, not 8 cm and 1000 samples/second. Obviously, high-speed visual tracking can be done. I wonder whether your budget is adequate for that. Also, there is a difference between being able to track something and being able to control it in real time.

How many spots will be on the airplane to identify it? What happens when a spot is not visible from the ground? If there are multiple identical spots, how will the camera know which one(s) are imaged? If the spots are not identical, how many pixels will you need to distinguish them? If the airplane is coming directly toward or away from you, how many spots will be visible?

John
 

Wouldn't it be easier to track an infrared beacon instead of painted spots? Preferably pulsed/modulated so the receivers don't lock onto birds or the sun.
 

...So an RF (Radio) system is also not possible...

Yes it is!
The company where I work achieves sub-cm indoor positioning at 2.4 GHz by carrier phase measurement: at 2.4 GHz, the wavelength is ~123 mm. A 1cm position change thus corresponds to an (easily detectable) ~30 degree phase change (multipath is another issue entirely, however...).

Attaining such accuracy over the 2 km distances you're interested in is non-trivial due to tropospheric effects though. Uncorrected (by pressure/temperature/humidity atmospheric models), these effects can contribute up to ~300ppm error (at 2.4 GHz).
 
To anticipate wrong conclusions, the said disturbances of atmospheric wave propagation affect optical measurements in the same order of magnitude.
 

You might also look into two sensing methods and a kalman filter to combine them.

this is actually an example kalman filter problem on wikipedia. the idea is to combine the short term accuracy of an accelerometer with the long term accuracy of GPS.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top