Re: [Flightgear-devel] GPS update rate

[Curtis O. <cur...@gm...>]

On Mon, Apr 14, 2008 at 11:29 AM, LeeE wrote:

> Oops - meant an exponential filter - seems to work ok, except for
> the inevitable glitch between 360/0 deg (I'm using
> indicated-track-true-deg) but then it resolves in 0.45 sec, which
> isn't too bad.  The lag affected a couple of controllers but they
> were easily re-tuned.


So how do you filter/average a heading?  I had to think about this on my
trip at sea for one of our applications.  I came up with a scheme that
actually hasn't been well tested; does anyone see any holes in this?

Let's say I want to do a simple moving average ... so the new value is
(let's say) 9 parts the previous filtered value + 1 part of the latest
sensor reading.  Doing that as a simple average though will glitch if your
values are coming in around 0/360.

So instead I looked at the (signed) angular difference between the current
filter value and the current sensor reading and normalized that to the -180
to +180 range.  Then the new filter value is computed as the old filter
value plus 1/10 of this difference.

I think that is equivalent.  But before I crash any airplanes, can anyone
poke any holes in my approach?

Thanks,

Curt.
-- 
Curtis Olson: http://baron.flightgear.org/~curt/