next topic »

[on:]

I have thought about an unusual technique for position tracking in 2D or 3D but I don't know if it's even possible.

The idea consists in inserting an XBee shield on a LilyPad carried by a dancer and use it as an emitter only. Then, I would place several receivers around the scene and connect them to another Arduino. If there's any way to measure the strength of the XBee signal from the emitter then I could track the position of the emitter.

What do you think about the idea ?

[Reply #1 on:]

Ciao,

You can take a look to an old post on Arduino Forum

http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1230326305

I think IR solution can give more accurate results than using XBee signal strength.
You can use

Other links
http://www.instructables.com/id/Wii-Remote-IR-Camera-Hack/
http://www.instructables.com/id/Wii-Remote-IR-Camera-Hack/
http://slopescope.co.uk/tech.html

Ciao,
Marco.

[Reply #2 on:]

What do you think about the idea ?

Not much. First, the signal strength name is RSSI - Relative signal strength indicator. That's like "on a scale of 1 to 5...". In other words, not very accurate.

Second, you'd need a boatload of XBees as receivers, connected to one computer. Not only would that be expensive, but you are limited by the number of serial ports that the computer can support, and the time it takes to poll each receiving XBee for it's relative signal strength indicator value.

Your refresh rate for knowing where the dancer was would be abysmally low.

[Reply #3 on:]

I think infrared has two drawbacks : first, it is very directive, so to track in two or three dimensions a bunch of sensors would be necessary. Second, it might be sensitive to lighting conditions, especially on a scene.

I understand the strength of the XBee's signal is coded only on a 5-steps scale, but why would a bunch of receivers be needed ? Couldn't one apply triangulation with only four receivers ?

Then, could Bluetooth or Wi-Fi have a finer scale to evaluate a signal's strength ?

[Reply #4 on:]

I'd suggest that you get two XBees. Program them to report RSSI values, and display the values on an LCD. Move one Arduino/XBee/XBee shield/LCD device around, and see how the RSSI values change.

This will tell you whether, or not, in your environment, the RSSI values means anything.

I understand the strength of the XBee's signal is coded only on a 5-steps scale,

I never said that. I said "like a...".

why would a bunch of receivers be needed ?

The worse the resolution, the more measurements you need to take to discern ANY difference.

Couldn't one apply triangulation with only four receivers ?

If all you are interested in is "in the same room", "in the same building", "in the same town", "in the same state" kinds of data, triangulation will work fine.

Then, could Bluetooth or Wi-Fi have a finer scale to evaluate a signal's strength ?

They are all radios, subject to exactly the same limitations. You have a portable radio? Take it outside and walk around the yard. As you move closer to, or farther from, the transmitter, does the signal change noticeably? No, of course not, unless your yard is many orders of magnitude larger than mine. The same problem exists with any device trying to measure signal strength. Strong enough to receive vs. not strong enough to receive is about the only measurement that is accurate.

[Reply #5 on:]

Wii Remote (as also Kinect) uses an IR camera and combined with IR light emitter make possible to analyze reflection of IR light by objects.
You can put reflectors on dancers.

An example of 3D depth sensors use made with two kinetic can be found here
Real-time People detection and tracking with multiple Kinect cameras

Detection is done from noisy depth measurements.

Another option is Xtion PRO from Asus
http://www.openni.org
Ciao,
Marco.

[Reply #6 on:]

Wii Remote and Kinect are very different animals. The Wii remote contains a 3D accelerometer and uses the infrared to calibrate one axis only, this IR sensor is much more primitive than the Kinect IR camera. The Wii remote doesn't have any sense of depth.

Anyway this example with two Kinects is very interesting. The Asus device seems very similar to Kinect. Does it have any specific asset ?

[Reply #7 on:]

Ciao,

Kinect and Asus Xtion PRO use the same tecnologies (developed by PrimeSense http://www.primesense.com/).
With Asus Xtion PRO You can develop also on Linux Ubuntu.

There is also a developers' kit from PrimeSense.
http://www.primesense.com/?p=514

For the Wii Remote only the IR camera hack is of interest for doing something like inspired to this http://naturalpoint.com/optitrack/products/motion-capture/.
With the dancer with passive IR reflectors (for the stage is simple to check how much IR light You have, use a camera + IR filter and check what You arrive to see or photograph (usually long exposure time is needed).

Here is a hack for Wii Remote IR camera http://letsmakerobots.com/node/7752.
I think some optical lenses are needed to extend range of the camera.

Ciao,
Marco.

[Reply #8 on:]

There is also this Truemotion 3D thing from Sixense which has been under development for quite a long time.

It seems it's going to be available soon trough Razer.

It is said to be magnetic but I wonder if it's not rather a kind of electric field sensing (like a Theremin).

[Reply #9 on:]

It seems I am not the only one to think about triangulation with XBees or similar radio devices.

http://www.sensor-networks.org/index.php?page=0827027001

[Reply #10 on:]

It seems I am not the only one to think about triangulation with XBees or similar radio devices

My reading of that link was that it is maybe of some use in a situation where you need to say "which room was s/he most recently in?", but not "which of the footlights is s/he nearest right NOW"

[Reply #11 on:]

PaulS is correct, you should get a couple of XBees and try the RSSI to see if the granularity is good enough for you.  I suspect the problems that PaulS went into will bring you to a halt on using them.  The RSSI doesn't seem to change until it almost goes away, if there is enough difference to be able to triangulate I would be surprised.  Also as Paul mentioned, the update time is probably much shorter than you want.  I suspect the person could move a couple of feet before you knew it.

[Reply #12 on:]

Indeed, other documents seem to confirm that this kind of tracking can be used with a lo-res granularity ("room identification") only. Too bad.