WildlifeResearcher
Posts: 2
Joined: Thu Mar 29, 2018 2:35 pm

Integrating ID with single-chip radar

Thu Mar 29, 2018 3:11 pm

I am a researcher investigating the movement and interactions of individual animals. I am looking for a solution that will allow me to track multiple individuals simultaneously within a defined area (~150 meters squared). RFID, while providing the ability to identify individuals, has far too short of a detection distance.

The use of short-chip radar seems like an interesting place to start, as it provides adequate detection range, but as far as I can tell it does not identify individuals. Could individuals be tagged with a small device that transmits an ID number in response to the radar signal?

I appreciate your thoughts.

User avatar
Ronaldlees
Posts: 294
Joined: Sat Apr 16, 2016 4:28 pm
Location: North Carolina, US
Contact: Website

Re: Integrating ID with single-chip radar

Mon Apr 02, 2018 2:34 pm

How difficult is it to detect radar signals? They're very short pulses, and detection may not be cheap enough. So, I'd think about light wavelengths instead. How many animals? How big are they? If they wore flags in an assortment of colors, you could use an Opencv cam and some software to pick them out. That wouldn't be good for very many individuals though. Are these animals candidates for facial recognition?

Maybe you could use a focused, scanning IR beam to trigger a device to send an ID. It would require a mechanical jig of some sort, and the scan-head output would have to be mapped to your 150x150 meter square area. It would have to be elevated, otherwise one animal could get in front of the other. Each device could be an arduino or some such, with a low power transmitter. That would be ~$30 per animal, which is not cheap. The IR would have to be low enough power to be safe, and wouldn't be 100 percent reliable.

The animals could wear cheap LED emitters that flashed a code, and then OpenCV or something similar could be used to catch the flashes. Maybe it'd need to be an IR camera and IR emitters. Filters are available that can pick up a bandwidth of only 40 nm. In any case, it might not be 100 percent reliable, due to ambient light conditions, but would be a stationary camera and a lot less mechanical trouble IMO. Probably it would not be very much $per-animal. The cam would have to be elevated, and the OpenCV would again have to be mapped to your space.

I've seen this question a few times. Never saw a very good answer though. Probably mine included.
I am the Umbrella man

WildlifeResearcher
Posts: 2
Joined: Thu Mar 29, 2018 2:35 pm

Re: Integrating ID with single-chip radar

Mon Apr 02, 2018 3:12 pm

Yeah, this could theoretically work with larger animals but I am unfortunately dealing with very small (<12 gram) birds so therefore the tag must be very small. Likewise, color is not ideal as it will probably affect many aspects of the bird's (or most any animal's) biology such as predator avoidance, prey capture, mate choice, etc.).

So, when you say wide-beam, do you mean that the signal may simply pass by the tag without detection? If so, then yes that would be a problem.

I've seen some RFID units that can detect as far as 10 meter away. Not sure what size of antenna is need though on the passive RFID tag. It might be possible to triangulate a bird's location with multiple units?
Ronaldlees wrote: A radar signal is probably too wide-beam to identify animals. So, I'd think about light wavelengths instead. How many animals? How big are they? If they wore flags in an assortment of colors, you could use an Opencv cam and some software to pick them out. That wouldn't be good for very many individuals though. Are these animals a candidate for facial recognition?

Maybe you could use a focused, scanning IR beam to trigger a device to send an ID. It would require a mechanical jig of some sort, and the scan-head output would have to be mapped to your 150x150 meter square area. It would have to be elevated, otherwise one animal could get in front of the other. Each device could be an arduino or some such, with a low power transmitter. That would be ~$30 per animal, which is not cheap.

The animals could wear cheap LED emitters that flashed a code, and then use OpenCV or similar to catch the flashes. That might not be 100 percent reliable, due to ambient light conditions, but would be a stationary camera and a lot less mechanical trouble IMO. Probably it would not be very much $per-animal. The cam would have to be elevated, and the OpenCV would again have to be mapped to your space.

I've seen this question a few times. Never saw a very good answer though. Probably mine included.

User avatar
Ronaldlees
Posts: 294
Joined: Sat Apr 16, 2016 4:28 pm
Location: North Carolina, US
Contact: Website

Re: Integrating ID with single-chip radar

Sat Apr 21, 2018 8:05 pm

I'm not an expert in radar, so this is a hobbyist's opinion, powered by being a boating/marine enthusiasts and ham radio guy. The ideas I'm tossing out here are probably worth what you paid for them ($0) :-) But, here goes anyway ...

The earlier automotive range finding chip radars (probably the only thing affordable) - had small parabolic fixed antennas because they needed to check the distance only in a certain direction. For a radar to "activate" a sensor on a tag, its beam would need to be steered. The sensor might report the ID when struck (although detecting the radar signal would not be trivial), and the radar could then maybe record the distance and position of the tag. The steerable antenna could be a parabolic antenna or a digitally steered antenna array, with the former being a mechanical nightmare when considering both horizontal and vertical planes, and the latter not uncomplicated with a half dozen antenna elements or more. In the case of the parabolic, the beam-width could be as small as 1 degree which might give good enough resolution, while the antenna array would give less resolution (5 degrees maybe?) - probably not good enough to activate the sensors of a single bird. All in all, the whole thing is probably a totally, totally, ridiculous thing to contemplate.

Like I said, I'd not seen any good answers to this kind of question, and your requirements (birds) make it even more difficult because they can work the x, y, and z axis coordinates, unlike for example a cow. And they're tiny. Great mind exercise to think about the problem though.

RFID has the same problems as the radar IMNEO (in my non expert opinion), in addition to not having any range capability. Technically, some tags have position via phase, but what I've read (below) says reflections may make the position unreliable:

https://www.quora.com/Can-the-position- ... r-3D-space

I don't know if that happens in many cases, so don't really know if it's an option. Maybe that article is similar to what you were looking at? For RFID and cows, one might plant tags in the pasture, and put readers on the cows. But then the dung could cover the tag, rendering it useless. This post may be dung.

Here is a bluetooth based position/tagging system, but it looks like only 2D, not the 3D you need:

http://quuppa.com
I am the Umbrella man

Return to “Automation, sensing and robotics”