IR Sensors- will they attack hot Chiaphuas?

[ErikJ]

Al pretty much nails it. The whole modulation thing can be tricky for anyone who hasn't seen it before, so that's probably the big cause of confusion (ie, IR frequency != 40KHz).

However, it should be noted that nearly all (I think it's all) TV/Stereo remotes operate at 40KHz (it's actually just a shade off). Most off the shelf commercial IR sensors (ie, the kind you buy at Radio Shak) only look for this 40KHz signal, and filter everything else out. The way your TV, and your FIRST robot, know the difference between the signals is in the coding scheme used on top of that 40KHz carrier. So... if you go off and write your own code, yes, TV remotes could be a source of interference. Or, perhaps cooler, you could probably allow (for demo/debug purposes) the FIRST controller to use a TV remote as an input, provided you knew the coding scheme of the controller (you can find many of these online).

Something I haven't seen made mention of yet is multipath interference*. I worked on an IR project a few years ago and would get strong multipath interefence when I was transmitting near large metal objects (in this case, lockers at the HS we were working with at the time). I'm interested to see if this becomes an issue, especially as one or more robots approaches the beacon. I'm not sure if this will be a major issue (I suspect not, but it could be depending on robot geometry), but the fix should be simple enough if it is.

* multipath interference - in short, when you recieve a signal that has been reflected off a few objects before you receive it. for example, in a car radio you have a direct path (radio tower to you), and a ground bounce path (tower, ground, you) - mutliple paths, hence multipath interference. In our example, this means that robot reflecting IR waves from the beacon appears to be a source, potentially causing problems for your location algorithm.

[Al Skierkiewicz]

Originally Posted by ErikJ

Something I haven't seen made mention of yet is multipath interference*.

* multipath interference - in short, when you recieve a signal that has been reflected off a few objects before you receive it.

Multipath is a degradation of signal due to a reflected signal arriving with the direct signal at a receiver. When the reflected siganl is far enough apart that the reflected signal is 180 degrees out of phase and near the same level as the direct signal, the two cancel out. Obviouosly at the frequency we are interested in (the IR frequency not the modulation 40kHz) there can be several objects on the field that will produce reflections that are 1/2 wavelength or multiples of 1/2 wavelength giving the out of phase condition.

[ErikJ]

I thought about that source of interference as well, but at these wavelengths very slight differences in distance will result in going from constructive to destructive interference (and back). I figured that then on average the chances of a missed detection are probably pretty slim. That brings my main concern back to the stray signal problem.

We have essentially a Direction of Arrival (DoA) problem - for more info just google for more info, or search IEEExplore if you have access. A similar problem has been solved many times over, especially in test ranges for antennas and radars. They have essentially the same problem we do - you can't distinguish a reflection off a wall from a reflection from the object. Now those guys use some pretty fancy techniques to eliminate those issues, but we can take a page from their notes. Since we have a priori knowledge of the source location (something I didn't have in the example I cited previously - in that case my beacon was moving), we can ignore signals coming from regions we don't expect to see on in. Now, this brings up some potential navigational pit falls, but if you used IR search/track in combination line following you'd be in good shape. If you're just randomly searching for a beacon then stray signals *could* become an issue. It's hard to say for sure (especially for me to say since I'm an EM guy not an E/O guy) until somone tries it out. However, since again the potential source of the stray signal is moving (quickly with respect to wavelength), this might go away just like the true multipath problem.