I am interested in somehow tracking the "compass" direction of a moving object from a reference-station -- in other words, something similar to submarine sonar systems (although I'm unsure how those determine the angle/direction of the target).

Note that I only need the direction/2D-angular-position of the object, and don't care about its actual distance/spatial-position. So, I wish to know, e.g., the object is at South 45° East from the reference-station.

Some notes:

  • Maximum size of circuitry at moving object: 10 cm X 10 cm
  • Maximum size of circuitry at reference station: 15 cm X 15 cm
  • The angular measurement needed is a relative one (i.e., as measured from the reference-station), not an absolute one.
  • An accuracy of +/- 10 degrees is more than good enough.
  • Measurement/tracking rate can be once per second, but faster wouldn't hurt!
  • While this tracking is being done, the moving object might be between 3 meters (min) to 100 meters (max) aerial distance away from the reference-station.
  • Assume that cost is not critical and that both the object and the reference-station can be electronically equipped (i.e., with sensors/receivers/transmitters/etc.)
  • The object and the reference-station will be located in an outdoor/urban environment.

What technique could I use to get started on achieving this kind of direction measurement/tracking?

I've considered using standard wireless signal transmission, but with this, I've only been able to measure RSSI (signal strength), and not the actual angle/directionality! I'm guessing I need some sort of directional transmission sent from the object, whether ultrasonic or RF, and correspondingly a 360-degree sensor array on the reference-station to determine direction.

  • \$\begingroup\$ The usual way of doing this by radio uses an omnidirectional antenna on the remote object, and a directional antenna array at the receiver. Search for information on "radio direction finder". \$\endgroup\$ – Peter Bennett Aug 14 '14 at 22:54
  • \$\begingroup\$ Do you need the direction of the object absolutely, or relative to the reference station? \$\endgroup\$ – shuckc Aug 14 '14 at 23:00
  • \$\begingroup\$ @shuckc: Relative to the reference station. I've edited the question with this info. \$\endgroup\$ – boardbite Aug 14 '14 at 23:06
  • \$\begingroup\$ Your question talks about RF solutions. Are line-of-sight solutions acceptable? \$\endgroup\$ – gbulmer Aug 27 '14 at 1:21
  • \$\begingroup\$ @gbulmer: Yes, line-of-sight is acceptable. Any thoughts? \$\endgroup\$ – boardbite Aug 27 '14 at 18:39

If you're locating an object through auditory means, you'll need at least two receptive transducers - measure the difference in sound arrival time. Three are much better; you won't have to rotate the transducers for a second "ping" arrival.

If you're locating through radio means, you'll need a rotatable directional antenna... or three omnidirectional radio antennae. If one, you'll rotate it until signal strength is strongest. If three, you'll select the strongest and extrapolate direction from the relative strengths received by the other two.


One last suggestion - you could put GPS on the vehicles and have it broadcast absolute position, then calculate the offset from the base station's GPS location in software. I'm sure you could mock it up with a couple of mobile phones and go more bespoke if it works.

Otherwise it's going to be radio direction finding, and you're going to need the help of a real electronics engineer for that ;-)

  • \$\begingroup\$ The GPS is the best shot, also because a 10° precision is needed. The only problem is that it will suck when the moving object is near the ground station. \$\endgroup\$ – Vladimir Cravero Aug 30 '14 at 8:50

(Update: Not suitable for the question, OP needs relative direction not absolute)

You can do this fairly well with an I2C compass module hooked up to a microprocessor on the object. This samples the compass direction and sends the heading to a serial port. You then use any off the shelf UART-to-wireless bridge to send it back to the reference station. You might get away with a Bluetooth module for 100m if it's the higher powered class II. For longer distances, especially if you don't have perfect line of sight, some of the FM band transmitter/receiver pairs might be more suitable, but you'll have to roll your own framing and reliability/checksum layers down that route.

If the object is a motorised vehicle of some sort, try and maximise the distance between the compass and the motorised actuators as their magnetic fields will (at best) skew your readings. Perhaps you can time readings when the actuators are off. I did something very similar for my graduate project.

  • \$\begingroup\$ Unfortunately, yes, relative direction is what I want, not absolute. But thanks for the suggestion on the FM transmitter/receiver idea. \$\endgroup\$ – boardbite Aug 14 '14 at 23:11
  • \$\begingroup\$ If you have many objects to track you might also want to try some IR LEDs on the vehicles and use 4 90deg field of view cameras back-to-back to locate them visually. It's not a trivial software problem (and needs calibrating) but is worth considering. \$\endgroup\$ – shuckc Aug 14 '14 at 23:22
  • \$\begingroup\$ I considered this (and the software won't be an issue since I'm willing to work on that!), but the issue with IR LEDs (especially for long-range) is that daylight will wash out the light, and also requires line of sight. \$\endgroup\$ – boardbite Aug 14 '14 at 23:23
  • \$\begingroup\$ If you've got the absolute direction of the base and the absolute direction of the object, getting their relative directions is as simple as substracting the angles! \$\endgroup\$ – Guillermo Prandi Aug 28 '14 at 22:58
  • \$\begingroup\$ erm, no it's not! OP needs the relative direction of the object as seen from the base, not the difference in angles. Put another way, the angle of the object is irrelevant. \$\endgroup\$ – shuckc Aug 29 '14 at 9:48

Personally, I think that RF is the way to go. You could possibly build a system similar to a lightening detector which has three antennas in a triangle shaped fashion. You can have a RF transmitter on the object that emits a signal every so often. These frequencies are detected by the three antennas at different times, and based off of the time difference and how strong the signal is you can basically tell how far away the object is along with its angle from the reference station.


Something that I thought of just off the top of my head is putting a wireless transmitter (RF or otherwise) and then the corresponding receiver at your base station but mounted on a small rotating rod or something, have a case around the receiving antenna so that it would only pick up a signal from a certain direction, ie. straight in front, but still allow it to receive from varying heights, so maybe a curved slit in a metal box/ dome around the receiver would simulate this more or less. Also the transmitter would not have to be directional, just powerful enough to transmit the distance you need it to and then some just to be sure it'll be picked up.

If you use a motor that you can track its own position so you would know which way the receiver is pointing at any given point and then when the signal being received as at its strongest/ peak, you can assume that that's the direction your object is and using the position of the motor you could easily work out the relative position of the object. Could fairly simply be implemented with a micro controller which takes the input from the receiver and can compare the received signals over one complete revolution of the rotating rod and find where the signal was strongest and then have it record this as the position.

Position could be worked out if the frequency/ period of the motor/ rotating rod is known as you would know that readings would be taken after each revolution and if you set a sampling rate of 360 times faster than the frequency you would get a reading for each degree around it's scan area and theoretically could get a pretty darn accurate position for the object.

Note: The signal strength could also be used to have a rough guess of how far it is, if you know the signal strength of it at x meters away then you could possibly work out the distance with the recieved signals using this value (assuming signal strength is linear with distance away, otherwise it could prove a little bit more tricky to do this)

  • \$\begingroup\$ Isn't this how 1st gen radar works? Other people are suggesting 2+ antennas to create an antenna array that would remove the need for moving parts. Admittedly motors are easier than calculating constructive and deconstructive waves. \$\endgroup\$ – lm317 Aug 28 '14 at 4:50
  • \$\begingroup\$ I suppose multiple antennae would be an easier model to create but harder to calculate the result from the readings, where as this method would be relatively easy to get a result from once set up. \$\endgroup\$ – MrPhooky Aug 28 '14 at 7:49

For such distances, and an urban setting, visible/IR emission or sound/ultrasound are probably out. I do not have any experience in the field myself, but maybe some keywords to search for will help you get started. I found a nice looking overview of the different technologies here: http://www.denisowski.org/Articles/Denisowski%20-%20Comparison%20of%20Radio%20Direction-Finding%20Technologies.pdf .

Another good intro to the field with lots of pictures: http://telekomunikacije.etf.bg.ac.rs/predmeti/ot3tm2/nastava/df.pdf

I'd say your best bet is looking for research articles and HOWTO guides on Adcock pair direction finding and Watson-Watt, or maybe the correlative interferometer direction finding method. If you have access to any academic/professional-society databases, you should try those first. Crossed ferrite loop antennas would probably fit your application given your size constraints.


I'd ping the remote device with an omni RF pulse from the local unit and have the remote device reply with a Xenon strobe which the local unit would pick up with a video camera.

The camera would rotate at a rate which would keep the remote in its field of view for at least two pings, assuring that it would receive at least one strobe in order to acquire the remote.

Once acquired, the camera would slew in the direction required to servo the strobe flash to its film plane's focal point and, depending on the tracking resolution required, adjust the ping rate accordingly.

Also, to minimize the effect of ambient light on the camera, a properly proportioned lens hood would probably be needed, and perhaps identical color filters over the strobe and in front of the camera's lens.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.