Some days ago I posted here a topic about filtering an IMU's data. Someone gave me a tutorial. I thought I understood after I read it. But I was wrong. I found it to be to basically explained.
I want to filter my 2 sensors (an accelerometer and a gyro) and simply I can't understand the theory. The Kalman filter, the complementary filter, FIR, IIR etc.
I've been documenting about them for days and days, I tried to understand their principle, how they work, but it's simply too hard for me. For example, I know that I have to fusion the sensors (at least 2), but I don't know the why, how?
I've barely succeed to make my own low-pass filter, which took me awhile to understand it.
I'm in the 10th grade, maybe this is the problem, and the kalman for instance, from which my dad told me, is being learned in the college.
Can someone just help me? How could I resolve this problem?
The simplest low pass filter suitable for noisy signals is "floating average" (I don't know whether this is the correct name, but I call it like that)
If you have variable "x" you want to filter, then you can use this code:
last = 0;
function timer()
{
last = x0.1 + last0.9;
}
By calling the timer() function, the variable "last" slowly converges to the "x" value, the sum of weights (0.1+0.9) must be equal to 1. By changing the ratio you also change the "speed" of filter.
FIR filters are complicated to design. If you want some specific filter with properly set cut off frequency and attenuation, you will need some filter designing tool (MATLAB has some toolboxes for this purpose). The IIR filters are not often used, because they are "crazy", by incorrect setting of the filter coefficients you can get some oscillating beast.