Path: Common/Estimation
% Interactive Multiple Models prediction step.
This does the interaction and filter steps.
The inputs and outputs only show data structure fields used in this
function. This assumes that the number and definitions of the state
do not change. There are n models and m states.
IMMUpdate is normally called after IMMPredict.
A complete IMM implementation is
IMMPredict( d )
IMMUpdate( d )
--------------------------------------------------------------------------
Form:
d = IMMPredict( d )
--------------------------------------------------------------------------
------
Inputs
------
d (1,1) IMM data structure
.use (1,1) 'KF' 'EKF' 'UKF' Only KF is implemented
.pIJ (n,n) Model transition probability
.muI (1,n) Probability of being in that model
.mI (m,n) Means for each model
.pI {n} Covariance matrix for each model
.aI {n} State transition matrix for each model
.qI {n} Plant noise matrix for each model
-------
Outputs
-------
d (1,1) IMM data structure
.cHatJ (1,n) Scaling factor
.mPI (m,n) Mean vectors for each model
.pPI {n} Covariance matrix for each model
--------------------------------------------------------------------------
Common: Estimation/KFPredict
Back to the Common Module page