Partially observable Markov decision processes (POMDPs) form an attractive and principled framework for agent planning under uncertainty. Point-based approximate techniques for POMDPs compute a policy based on a finite set of points collected in advance from the agents belief space. We present a randomized point-based value iteration algorithm called Perseus. The algorithm performs approximate value backup stages, ensuring that in each backup stage the value of each point in the belief set is improved; the key observation is that a single backup may improve the value of many belief points. Contrary to other point-based methods, Perseus backs up only a (randomly selected) subset of points in the belief set, sufficient for improving the value of each belief point in the set. We show how the same idea can be extended to dealing with continuous action spaces. Experimental results show the potential of Perseus in large scale POMDP problems.

Partially observable Markov decision processes (POMDPs) form an attractive and principled framework for agent planning under uncertainty. Point-based approximate techniques for POMDPs compute a policy based on a finite set of points collected in advance from the agent's belief space. We present a randomized point-based value iteration algorithm called Perseus. The algorithm performs approximate value backup stages, ensuring that in each backup stage the value of each point in the belief set is improved; the key observation is that a single backup may improve the value of many belief points. Contrary to other point-based methods, Perseus backs up only a (randomly selected) subset of points in the belief set, sufficient for improving the value of each belief point in the set. We show how the same idea can be extended to dealing with continuous action spaces. Experimental results show the potential of Perseus in large scale POMDP problems.

Spaan, Matthijs T. J. (Instituto Superior Técnico) | Lima, Pedro U. (Instituto Superior Técnico)

Nowadays many urban areas have been equipped with networks of surveillance cameras, which can be used for automatic localization and tracking of people. However, given the large resource demands of imaging sensors in terms of bandwidth and computing power, processing the image streams of all cameras simultaneously might not be feasible. In this paper, we consider the problem of dynamical sensor selection based on user-defined objectives, such as maximizing coverage or improved localization uncertainty. We propose a decision-theoretic approach modeled as a POMDP, which selects k sensors to consider in the next time frame, incorporating all observations made in the past. We show how, by changing the POMDP's reward function, we can change the system's behavior in a straightforward manner, fulfilling the user's chosen objective. We successfully apply our techniques to a network of 10 cameras.

Egorov, Maxim (Stanford University) | Kochenderfer, Mykel J. (Stanford University) | Uudmae, Jaak J. (Stanford University)

This paper introduces an extension of the target surveillance problem in which the surveillance agent is exposed to an adversarial ballistic threat. The problem is formulated as a mixed observability Markov decision process (MOMDP), which is a factored variant of the partially observable Markov decision process, to account for state and dynamic uncertainties. The control policy resulting from solving the MOMDP aims to optimize the frequency of target observations and minimize exposure to the ballistic threat. The adversary’s behavior is modeled with a level-k policy, which is used to construct the state transition of the MOMDP. The approach is empirically evaluated against a MOMDP adversary and against a human opponent in a target surveillance computer game. The empirical results demonstrate that, on average, level 3 MOMDP policies outperform lower level reasoning policies as well as human players.