Re: Kalman Filter to recover the Accelerometer Bias



Perhaps it is time to stop consulting engineering types, who, after all,

can only do the physically possible, and go looking for magicians?


Alright, i guess i wasn't clear or you didn't understand my problem. What
i am trying to do is simulate what i will get from an Accelerometer which
say has scale,bias error + noise. I was trying to do reverse ENGINEERING to
see if kalman filter will identify and remove the added bias and scale
factor.

For the simulation, i have a distance measure ( from a distance sensor i
have now but won't have in the future when i only have accelerometer), so i
differentiated w.r.t twice to get acceleration, then added scale factor and
a bias + assumed noise as the variance of this simulated scale factor and
bias added acceleration to make it similar to what i will get from an
accelerometer which has white noise .

Now i have a simulated acceleration measurement which has scale and bias
and i want to use kalman filter to remove them, isn't that still
engineering? The measurements are the simulated measurements.

hence state vector x = [ acc
bias]
my h(x) = scaleFactor * acc + bias
my H matrix the partial differentiation would be
H = dh(x)/dx = [ scale Factor 0]
A = [1 1
0 1]
R = var( simulated acc)
Q = some value lesser than R

Starting Values of State vector = [0
sensor's bias from the datasheet]
starting covariance Psub0 = Q

Why can't i do this?





.

Relevant Pages

  • Re: Accelerometer and Kalman Filter
    ... Here is why i think i can use a kalman filter. ... going to just based on accelerometer output, that would mean i just have ... removing the scale error and for simplicity sake assume acc_noise = 0. ... Now the measurements i make is a acceleration measurements with a bias ...
    (comp.dsp)
  • Re: Confidence Intervals for Bridge Simulations (long)
    ... A 1000 run simulation that gave the same 41% ... You have neglected to take into account the bias of the sample. ... the Dealmaster Pro/Deep Finesse double dummy analysis ... had NEITHER 3 key cards NOR 2 key cards plus the QS. ...
    (rec.games.bridge)
  • Re: COBRA
    ... I'm in the process of writing a "Simulation Primer" for the website to ... detail the various ways and purposes of simulation work, ... I've come to recognize, however, that DD bias ...
    (rec.games.bridge)
  • Re: Confidence Intervals for Bridge Simulations (long)
    ... A 1000 run simulation that gave the same 41% ... of accuracy would be sufficient for most bridge purposes. ... You have neglected to take into account the bias of the sample. ...
    (rec.games.bridge)
  • Re: best software environment for numerical analysis
    ... I have another "problem" that is purely statistical, not a simulation. ... A randomizing machine has a bias, and from a sample of "spins" i want ...
    (sci.math.num-analysis)