KFPredict:

--------------------------------------------------------------------------
   Linear Kalman Filter prediction step.
   This assumes a discrete model of the form:

   x[k] = a[k-1]x[k-1] + b[k-1]u[k-1] + q
   y[k] = h[k]x[k] + r

   b and u are optional.

   If only one argument is entered it assumes it is a datastructure of
   the form d = struct('m',m,'p',','a',a,'q',q,'b',b,'u',u)
   or  d = struct('m',m,'p',','a',a,'q',q,'bU',b*u)

   Since version 11.

--------------------------------------------------------------------------
   Form:
   [m, p] = KFPredict( d )
   [m, p] = KFPredict( m, p, a, q, b*u )
   [m, p] = KFPredict( m, p, a, q, b, u );
--------------------------------------------------------------------------

   ------
   Inputs
   ------
   m             (n,1)  Mean
   p             (n,n)  Covariance matrix
   a             (n,n)  State transition matrix
   q             (n,n)  Model noise matrix
   b             (n,p)  Deterministic input matrix
   u             (p,1)  Deterministic input vector

   -------
   Outputs
   -------
   m             (n,1)  Mean
   p             (n,n)  Covariance matrix

--------------------------------------------------------------------------