Directional tracking with a range of ~200' ...RF or alternatives?

Goal: Track a person with a wearable device (say, cell phone size max) in an indoor setting, with assorted objects which confound the existing vision based systems for keeping a phone/video camera aimed at an individual.

My research has shown that RSSI is unlikely to be available from the cheaper 433 mhz tx/rx pairs, and even if it was, it’s a poor indicator of proximity, or in an instance of a directional antenna, if it’s aimed at the tx. The directional antennas also tend to be bulky.

Preferably this would be done with just one base station, not a triangulation setup.

Would prefer to avoid IR or ultrasonic beaon type approaches.

Any additional architecture options come to mind?

Use the same device as wild life people do when tracking animals in the wilderness. Calls for a quite large tracking antena.

Any directional antenna will have side-lobes that also give an indication of direction, but less signal than the main lobe of the antenna. There are antenna systems that will give you the actual direction, but are way beyond what an Arduino will be able to do and way beyond what you can afford.

Paul

Paul_KD7HB:
Any directional antenna will have side-lobes that also give an indication of direction, but less signal than the main lobe of the antenna. There are antenna systems that will give you the actual direction, but are way beyond what an Arduino will be able to do and way beyond what you can afford.

Paul

Looking at the patterns for typical yagi antennas, it seems that the gain is at least 10db better at front center vs the other lobes, which seems like something where if the antenna can be set on an appropriate horizontal plane, and is free to scan the area for the highest signal strength, it could be treated somewhat like a line tracking robot.

Maybe three of these?

At $30 each, it's the cheapest solution I see.

You can get a commercial system with four beacons from MarvelMind for $400.

You can get a pre-made "Move and See" tracking system from B&H Camera for $835.

johnwasser:
Maybe three of these?
Ai-thinker® wideband indoor positioning module close-range high-precision distance measurement nodemcu-bu01 development board Sale - Banggood.com

At $30 each, it's the cheapest solution I see.

You can get a commercial system with four beacons from MarvelMind for $400.

You can get a pre-made "Move and See" tracking system from B&H Camera for $835.

Thanks John!
I'm starting to dive into the bu01 datasheets, had to get into the 245 page version to find max range! With 60-200m of range possible, this is seeming like a solid lead. Time to do some trig based on their accuracy data and see how little spread is needed for good-enough triangulation.

I've used IR TOF in sensor applications before, but never RF. Guess it's a logical approach given that it's basically GPS on a small scale.

I was aware of the ~$800 options, but it crosses the $ pain threshold in a way that spending many more development hours doesn't.

xxguitarist:
Looking at the patterns for typical yagi antennas, it seems that the gain is at least 10db better at front center vs the other lobes, which seems like something where if the antenna can be set on an appropriate horizontal plane, and is free to scan the area for the highest signal strength, it could be treated somewhat like a line tracking robot.

...except for a number of spoilers:

  • an RF emitter creates numerous reflections off all the conductive surfaces around it. So you will pick up all the signals from the reflections, they are off the beam axis, even behind you sometimes

  • the more directional a yagi is, the more elements it must have, which makes it very awkward, especially under 1GHz where the wavelengths are substantially long. Even a 3 element yagi at 433 MHz is quite large

  • gain is relative, so you need to compare front/back/side readings to get a bearing

-a yagi is sensitive to conductive materials around it, so must be physically isolated to some extent (that is why TV antennas are on a mast, you can't just attach them willy nilly to a wall or tower frame

I've probably missed a few more...

The accuracy of a GPS is some 10 meters, 30 feet, unless You have additional capacity available and paying for it. Then 1 meter, 3 feet is possible.

Railroader:
The accuracy of a GPS is some 10 meters, 30 feet, unless You have additional capacity available and paying for it. Then 1 meter, 3 feet is possible.

but not indoors

aarg:
but not indoors

Being indoors is not favourable. Depending on the building, it's impossible, or possible.
Started my GPS, unpowered for a year, inside the building. It took 6 minutes to find more than 3 satellites. Making a fast walk, speed indication works.

If you have a camera focused on the subject, use machine vision.

Vision systems calls for fat wallets...

xxguitarist:
Looking at the patterns for typical yagi antennas, it seems that the gain is at least 10db better at front center vs the other lobes, which seems like something where if the antenna can be set on an appropriate horizontal plane, and is free to scan the area for the highest signal strength, it could be treated somewhat like a line tracking robot.

All true. But to do what you want the object you are tracking needs to be 4 to 5 wavelengths away from the antenna.
Paul

Railroader:
Started my GPS, unpowered for a year, inside the building. It took 6 minutes to find more than 3 satellites.

What was the position accuracy ?

srnet:
What was the position accuracy ?

Can't tell as I use only height, speed, direction, time, date, VDop, HDop and the number of satellites. After 3 minutes a 3, 4 ,5 satellites were found and later it reached 11 - 12. Wouldn't that tell that a position will be as good as anything?

I think that John nailed it with the Decawave DW1000 based dev board. It's basically a local area TOF/ self contained GPS system. It'll need two base stations, I haven't had the time yet to do the trig.

Railroader:
Can't tell as I use only height, speed, direction, time, date, VDop, HDop and the number of satellites. After 3 minutes a 3, 4 ,5 satellites were found and later it reached 11 - 12. Wouldn't that tell that a position will be as good as anything?

The only way I know of to check the actual position accuracy of a GPS is to compare the position the GPS is reporting with the actual location, which is not very difficult.

Rereading the question from OP I think that GPS is not the solution anyway, whatever the number of satellites found.