Prediction

   Belief evolution given a transition model.
  
   Updates a particle-based belief with a given transiton probability function, 't'.
   Function 't' is the outcome of GetActionModelFixedA for the corresponding
   action model.

   Note that the prediction and update steps using particles are tied
   together (see Section 5.3.2 of the paper). However, for compatibility
   with the rest of belief representations it is nice to have separate 
   Prediction and Update functions.
   Here we store a backup of the particles (to use them in the Update) and
   we randomly modify them using the action model.

0001 function bOut=Prediction(b,t,Sp)
0002 %   Belief evolution given a transition model.
0003 %
0004 %   Updates a particle-based belief with a given transiton probability function, 't'.
0005 %   Function 't' is the outcome of GetActionModelFixedA for the corresponding
0006 %   action model.
0007 %
0008 %   Note that the prediction and update steps using particles are tied
0009 %   together (see Section 5.3.2 of the paper). However, for compatibility
0010 %   with the rest of belief representations it is nice to have separate
0011 %   Prediction and Update functions.
0012 %   Here we store a backup of the particles (to use them in the Update) and
0013 %   we randomly modify them using the action model.
0014 
0015   bOut=b;
0016   
0017   bOut.noiselessMovedSamples=bOut.samples+repmat(get(t,'mean'),1,bOut.np);
0018   bOut.noise=get(t,'covariance');
0019   for i=1:b.np
0020     bOut.samples(:,i)=rand(t+bOut.samples(:,i));
0021   end
0022   
0023   
0024