Quote Originally Posted by wasssup1990 View Post
Ok I got the timer going, but brace yourself for a possible kalman filter question. I havn't found a single website that can explain kalman filters in laymens terms.
And you probably won't find one.
Best explanation I can offer is that a Kalman filter makes a best 'future guess' based on the accuracy of it's past 'future guesses'.
If the Kalman filter isn't sure of itself, it's 'future guess' goes out the window.
I remember this from working the INU's on -130's. Ring Laser Gyro based INU's. The aircrews wouldn't believe the INU's position, so they'd update the things fairly often from TACAN or RADAR or whatever, even handheld GPS's. Then they'd whine when they got back because the INU would start drifting really badly inflight. The crews would come back and whine, we'd explain to them that they had ruined the INU's confidence by telling the INU it wasn't where it thought it was even though the difference between where the INU thought it was and where the crew told it that it was was very small. But of course, the aircrews knew better, because they're officers, and have degrees right?
We'd have to go out and run a ground align on the INU about a half a dozen times and each time, just let the INU cook by itself for awhile to build up it's confidence. At first it would sit there and drift away from the spot the aircraft was on. After a couple of time, the drifting would get less and less and eventually, it wouldn't drift at all.
But all that explanation doesn't help a person write any code now does it?