In this video, I am using a UMI, a sonar for altitude, and a wii-remote camera. The wii-remote camera is a very small camera that is included in the wii-remote.
The wii-remote camera gives directly the 2D coordinates of the IR spots in the image. It's interface is I2C. This is really a great sensor!
On my UAV, the wii-remote camera looks under.
I put IR LEDs on the floor.
I have put some lenses on this wii-remote camera, because the field of view was too narrow.
I still plan to make my drone navigate autonomously in an indoor environment, with several IR spots put on the floor.
For french-speeking people, you can see all the work that I have done on this web-site.
I will soon translate it in english.
Do not hesitate to ask me any question about this project.
Leon.
Comments
@ Roald:
Sorry, but I don't have the reference of the lenses anymore. I have totally disassambled the 2 lenses that I used.
@ Randy : Thank you for this idea. The assembly with "hot glue" works fine now. I don't want to use a more complex solution for the moment.
@ agmatthews : I think that it would be more robust to use :
* A sonar for altitude
* compass for orientation
* Wii camera for x-y position
Using just a wii camera for all purposes (including orientation and altitude estimation) will reduce the field in which the indoor-UAV can fly.
I plan to use several IR spots spread at different position on the ground. With this, I think that I can make my UAV fly autonomously in a wide area.
Leon.
If you want to get more sophisticated with mounting the lense, you might try making a custom lens mount using a 3d printer (or a 3rd party 3d printing service like shapways.com). I've done a lot of investigation of lenses and lens mounts for an unrelated (but in some ways similar) project. Geoff Barrows (a true master of robots + optics) gave me this suggestion and I've actually got an order coming from shapeways in the next few days. I'm not sure if the 3d printing gives enough detail to make the screw part of the mount but I'll find out soon.
When you look throuh the lenses assembly (with your eye), the image must be clear and wider than naturally.
And yes, the 2 lenses are just fixed with hot glue!
Leon.
@ Leon, I understand its one camera. I like your approach of having all the processing done on the copter. I had an idea of using a wiimote camera to replicate motion capture systems but have not had enough time. Keep it up.
> Matt said : "This is similar technology used in motion capture systems"
Yes, but this only uses ONE camera, and the camera is ONBOARD. All the computing power is inside the UAV. I really prefer this approach.
I am using a custom hardware, based on an "Embedded Master" (GHI Electronics). Not an "ardupilot".
The wii sensor can be interfaced quite easily. The interface is I2C (slave of course) and power supply is 3.3V. I think that the I2C interface is NOT 5V tolerant. It also requires a "clock" signal at 25Mhz, and a "reset signal".
The camera can detect up to 4 IR spots at the same time, and can refresh information at 200Hz!
I found information about this sensor on the folowing web sites:
http://wiki.wiimoteproject.com/IR_Sensor
http://procrastineering.blogspot.com/2008/09/working-with-pixart-ca...
http://www.stephenhobley.com/blog/2009/02/22/pixart-sensor-and-ardu...
I think that with those information, you probably could implement that on the "arducopter".
Leon.
great stuff. Could you post more information (or the code) you used to communciated with the wii remote's camera? Maybe you could also say what pins do what? Is it 3.3v or 5v?
the camera must have a little cpu on the back which is doing the image processing...I wonder what it uses.
again, great stuff. Looking at your website you've got a nice custom board there.
This is similar technology used in motion capture systems. Congrats
I forgot the scheme for better understanding: