Hi all,

I am pretty sure that you can determine a fair approximation of the orientation of a plane from GPS, 3 accelerometers and 3 gyros, although you might have to incorporate a model of the plane and use GPS, accelerometer and gyro information to get an accurate estimate of bank angle. Pitch and yaw should be no problem in any case. The GPS that I am recommending provides a very accurate 3D velocity vector based on Dopler shift. Together with gyro, accelerometer information, and a model of the plane, you can make a very good estimate of the orientation of your plane. The only caveate is that since GPS gives course over ground, if for some reason you prefer to navigate and control the plane with respect to the air instead of the ground, you will have to estimate wind velocity some how. Anyway, my goal for the spring is to perform aerobatics under autonomous control. I'll let you know how it works out. I will publish the firwmare.

With my previous board that had GPS, and only 2 accelerometers, and 2 gyros, I was able to construct a fairly accurate model-based (I had to include actual rudder and elevator deflections in the algorithm) estimator of yaw, pitch, and bank of a sailplane. I used it this past summer to provide pitch stabilization and a come-home feature for my Gentle Lady. Performance was very smooth with no problems during 100s of flights. The source code on the DIYdrones website is a simplified (I finally decided that I got better performance without bothering with mixing and bank angle) port of this firmware from the previous board to the new one.

You might be interested in reading some of the material I wrote on the subject of control and navigation algorithms, its available at

www.sparkfun.com on the page for the first version of my board, which by the way is now on sale for half price. If you are considering a sailplane, you might do well to buy the previous version of the board. The code for it is in assembly, but it is my understanding that there is a free C compiler available from MicroChip that is compatible with the 18F2520 PIC that is used on the board, if you have any interest in rewriting the firmware in C.

I invite anyone who wants to, to get in touch with me here.

Best wishes to all who read this.

Bill Premerlani