Vision Tracking Lighting Idea

I just read the FRC PDF on the 2012 vision tracking system. It talks about different lighting techniques that I found useful, however I had a question that is open for talking through. Instead of using regular LED’s around the camera for the retro reflective tape, why not use IR LED’s? Digital cameras easily pick it up, however that does bring up the question of “making the camera blind” so it can only see IR waves. I think was wondering if this were possible since it would cut down on a ton of processing power for an image. What does every one else think?

PS: If you’re wondering about edge detection in a language other than LV look up using a high-pass filter and a sobel gradient filter. These are used for the core of edge detection.

Using IR LEDs and an IR filter* over the lens sounds like a good idea. Note that visible LEDs tend to be brighter than IR LEDs, so experiment!

*These filters look black. Look on an IR remote control for an example. They allow IR though and block visible light. Many cameras have a filter that blocks IR light and allows visible through.

There one gotcha that you should test for first. Both CMOS and CCD based cameras are highly sensitive to IR and so they almost always contain a filter on them to block IR from their input. You should make certain the LEDs you choose can be seen by the camera at all before you take it too far.


If you do a search, you will find several discussions in threads about IR and the webcam. Specifically, the webcams we have have an integrated film on the lens that is not removable. You will have to replace the lens of the camera to do any work with IR.


Could you elaborate on this? Like actually placing the kinect ON the robot and using its camera. If yes then could you please tell us how my friend and I want to program this into our robot. Thank You :smiley:

I’d caution against using IR.

It may work in your shop or in practice, but the real headaches will occur when you go to competitions. Depending on the venue, arenas may have IR heavy lighting that may interfere with your own lighting.
I recommend using some very bright color instead. Same amount of work to get it working in software and less amount of work on hardware.

My $.02,
Jeremy G.

So…the Kinect projects an Infrared “beam” into the scene and depending on which driver/SDK you use(OpenNI, libfreenect, Microsoft SDK), you can get the IR feed directly from the Kinect. This feed is basically a grayscale image that represents the intensity of the Infrared received by the Kinect. Cool things happen with retroflective material (like that used on the Rebound Rumble targets). As for putting the Kinect on the robot…I dunno…Y U NOT KNOW!!!

Yes, and it should be just as simple as plugging the Kinect into the USB port on your cRIO… :smiley:

Exactly. Should be cake. :wink:

You are allowed to put a laptop on your robot, so theoretically, you could put a kinect and a small netbook on your robot, then connect it to the bridge and run a labview VI to send the information to your cRio.

Or even just use one of these on your robot, and use the USB host and Ethernet interface to talk to the cRio, that would be lighter, but much harder to accomplish.