Vlassis, Nikos[University of Luxembourg > Luxembourg Centre for Systems Biomedicine (LCSB) > >]

2005

Proc. Robotics: Science and Systems

217-224

Yes

Proc. Robotics: Science and Systems

2005

[en] We present a value iteration algorithm for learning to act in Partially Observable Markov Decision Processes (POMDPs) with continuous state spaces. Mainstream POMDP research focuses on the discrete case and this complicates its application to, e.g., robotic problems that are naturally modeled using continuous state spaces. The main difficulty in defining a (belief-based) POMDP in a continuous state space is that expected values over states must be defined using integrals that, in general, cannot be computed in closed from. In this paper, we first show that the optimal finite-horizon value function over the continuous infinite-dimensional POMDP belief space is piecewise linear and convex, and is defined by a finite set of supporting α-functions that are analogous to the α-vectors (hyperplanes) defining the value function of a discrete-state POMDP. Second, we show that, for a fairly general class of POMDP models in which all functions of interest are modeled by Gaussian mixtures, all belief updates and value iteration backups can be carried out analytically and exact. A crucial difference with respect to the α-vectors of the discrete case is that, in the continuous case, the α-functions will typically grow in complexity (e.g., in the number of components) in each value iteration. Finally, we demonstrate PERSEUS, our previously proposed randomized point-based value iteration algorithm, in a simple robot planning problem with a continuous domain, where encouraging results are observed.