jbw

I chose many of the components (motors/escs/props) based off of the Aeroquad project. I use a Xbee serial radio module to send tilt/heading/control gains. I wrote the gyro/accelerometer/magnetometer drivers, serial commands, PID control, and orientation filter. After a while of trying to write my own quaternion filter, I came across an efficient version I found in C and just converted to C#.

The top video is using Atan tilt approximation with a kalman filter to stabilize, the bottom using the more robust quaternion filter.

jbw, that's really neat - you mean you built the whole thing yourself, i.e. the bodywork, rc receiver module, motor controllers, etc or did you build onto existing q-copter product?

Bonus points for a video of the newer more stable version

One of the first flights since reworking the orientation algorithm.
I really only tested a few gains since the change, so there is some improvement to be done, but it is already 100x easier to control!

At one point I show it yawing, I then release and it returns to it's original angle.

All code is still managed!

Very nice! Is the improvement primarily because of the new motors? Or maybe the higher moment from the socks makes the system more stable for lower sampling rates

Bonus points for a video of the newer more stable version

Now that I can't do immediately. I'm waiting on more propellers after the power supply came loose and it boomeranged a trashcan.

Sorry about that, I'll edit it when I get the chance...

Unprofessional camera man...

Very impressive, jbw. I can only imagine what you can do with a bit of native code in there too...

BTW, any chance you can post a version without the last few seconds of audio? We'd like to tweet about it and show off your project, but don't want to have to add a language disclaimer

Bonus points for a video of the newer more stable version

Chris