Hello, I'm currently working on a project involving the tracking of a quadcoptor (a helicopter with four propellors.) I've taken a small remote control quadcoptor and used and arduino to replace the transmitter. I have not rewired the quad. I have replaced the transmitter so that i can control the quad from the computer.
I've been trying to use the Microsoft Kinect to track the quad but that seems to be much more difficult that i anticipated. I would like to be able to track the quad so i know exactly where it is in the room. Then i'd like to set coordinates and have it fly and stay at that position. What i'm having trouble with is figuring out the best method to track the quad in all three dimensions, besides adding sensors onto the quad as this would involve replacing the entire main board and making it too heavy.
I suspect that even with devices installed in the quadcopter what you want to do will be difficult if not impossible with DIY resources.
I don't think a GPS receiver in the 'copter would give you a sufficiently accurate position (if it worked at all indoors) and that is probably the easiest solution.
The alternative is some sort of radar/sonar system. But how would a "ground based" device "recognize" the quadcopter in order to measure how far away it is?
I've been trying to use the Microsoft Kinect to track the quad but that seems to be much more difficult that i anticipated
The Kinect is way more powerful than an Arduino, so a possible solution would be multiple coordinated Kinects,, like the ball location systems used in sports.
Using USB cameras you could use Roborealm to track a bright IR beacon (or any distinct color LED) attached on the quad. Two cameras would allow you to track it in 3D space.
You might make a modulated IR beacon for the flying object and use two IR detectors at right angles to each other being swept back and fourth by two servos (like sweeping radar antennas).
Then you should look at OpenCV. An IR LED on the copter will show up very brightly in the picture taken by a digital camera; OpenCV will help you do the image capture and some basic image manipulation to get the LED to stand out well in the picture. Then it's just a matter of identifying the brightest pixels in the image.
If you had a camera mounted on the copter you could point it at the ceiling where you'd have fiducial markers placed. reacTIVision will identify the x/y coordinates of the fiducial markers in the frame, and with a little trig you can get your 3D position.
I don't have the link to hand, but I've seen some very impressive demonstrations of quadrotor control in an enclosed space using 9DF sensing on each vehicle for basic flight control, and external fixed cameras picking up beacons on each unit to determine absolute positions to give precise control over the flight path. This seemed to give positioning down to accuracy of a couple of cm and enabled the drones to carry out complex manoeuvers very close to each other and to other objects. It looked as if it was the product of a substantial amount of effort over several years by a team of very smart people, and I think this would be very difficult to get working from scratch.
I haven't seen anyone doing it yet, but it seems to me that the IR position sensing technology being developed for CastAR would work very well as an onboard position sensor, so that the drone could work out its position and orientation relative to fixed beacons in the environment. Since the complex part of the beacon detection and geometry analysis is being done in hardware, there would be almost no overhead for this.
Thanks so much for the replies. I think what im going to attempt to do is add an IR Led onto the drone and tack it using the Kinect. That way I can still have depth of the room while tracking the drone. I'll Try to post back to let you guys know whether or not i get it to work!