Subtracting centrifugal acceleration from acceleration caused by movement

[sparkk]

I've been playing around with dead reckoning stuff on an rc plane by attaching a gps, 3 axis gyroscope, and 3 axis accelerometer to the plane.

When the plane isn't turning, my algorithm works pretty well however when I enter a turn, the readings get way off. Specifically, the plane appears to be moving at a faster velocity than it actually is thus all turns are overshot.

I believe I see this due to centrifugal acceleration and would like to cancel it out, but I'm getting quite lost, especially with 3d vectors. How might I go about subtracting this observed acceleration from acceleration caused by movement?

Thank you

 

[sparkk]

Let me add what I have identified so far.. for at least identifying what I believe to be the acceleration component I need to subtract:

F = ma_{c} = (mv^{2})/ r = mrw^{2}

The definitions are standard as of this wiki page:
http://en.wikipedia.org/wiki/Centripetal_force

I know:
(gps) speed = v
(gyroscope) angular velocity = w

I can calculate r via:
(mv^{2})/ r = mrw^{2}
v^{2}/ r = rw^{2}
v^{2} = r^{2}w^{2}
r^{2} = v^{2} / w^{2}
r = \sqrt{v^{2} / w^{2}}


With r and w known, I can now calculate a_{c} as follows:
ma_{c} = mrw^{2}
a_{c} = rw^{2}

Am I anywhere close to being on the right track?