(Somewhat) Useful for Localization in Topological Maps a p(x2x1a)=.9 p(xiX, a)=.05 X4:p(x4x1a)=05 Observations can be features such as corridor features Junction features, etc Belief Tracking Estimating p(x)is now easy After each action a, and observation zt VX∈X, update: ∑p( p This algorithm is quadratic in XI (Recall that Kalman Filter is quadratic in number of state features Continuous x means infinite number of states.(Somewhat) Useful for Localization in Topological Maps x1 x2: p(x2|x1,a)= .9 x3: p(x3|x1,a)=.05 x4: p(x4|x1,a)=.05 Observations can be features such as corridor features, junction features, etc. ● Estimating pt (x) is now easy ● After each action at and observation zt , ∀x∈X, update : ● This algorithm is quadratic in |X|. ● number of state features. Continuous X means infinite number of states.) Belief Tracking )()',|()|()( 1 ' xxp xpx t X t = t ∑ t − (Recall that Kalman Filter is quadratic in a x p z p