I'm building quadcopter and most of the control systems use one accelerometer and gyro. I've read few papers and usually accelerometer is used as reference to the ground because gyro slowly drifts away in time. But if quadcopter does some crazy maneuvering when force direction from accelerometer does not have to point to the ground than accelerometer data is useless. As well there is problem with centripetal force if the accelerometer is not directly in the centor of mass.
I was thinking about using multiple accelerometers. To fully determine position and motion of quadcopter one would need three accelerometers(If I have done the math right). This would kind of solve the problem with centripetal force
So I would like to know if anyone tried to use multiple accelerometers for better orientation estimation.