Institut de Robòtica i Informàtica Industrial

InformationGain

PURPOSE ^

Information gain by closing a loop with a previous pose.

SYNOPSIS ^

function ig=InformationGain(F,R,step)

DESCRIPTION ^

 Information gain by closing a loop with a previous pose.

 Computes the information gain of stablishing a link between the current
 robot's and that at time 'step' with a sensor with noise R.

 We take the natural logarithm and, thus, the output is given in nats.
 Just use log2 to the it in bits. If you do so you will need to
 adjust the information gain parameters accordingly.

CROSS-REFERENCE INFORMATION ^

This function calls: This function is called by:
  • Simulation Simulates a robot performing Pose SLAM.

SOURCE CODE ^

0001 function ig=InformationGain(F,R,step)
0002 % Information gain by closing a loop with a previous pose.
0003 %
0004 % Computes the information gain of stablishing a link between the current
0005 % robot's and that at time 'step' with a sensor with noise R.
0006 %
0007 % We take the natural logarithm and, thus, the output is given in nats.
0008 % Just use log2 to the it in bits. If you do so you will need to
0009 % adjust the information gain parameters accordingly.
0010 
0011   MarginalSigma=GetMarginalCovariance(F,step);
0012   
0013   PoseN=Pose(GetPoseMean(F,F.nSteps));
0014   PoseI=Pose(GetPoseMean(F,step));
0015   
0016   [Hn Hi]=Absolute2RelativeJacobian(PoseN,PoseI);
0017   
0018   S=R+[Hn Hi]*MarginalSigma*[Hn Hi]';
0019   
0020   ig=0.5*log(det(S)/det(R));
0021


Institut de Robòtica i Informàtica Industrial

Generated on Fri 24-Jul-2009 12:32:50 by m2html © 2003