Perseus: Randomized Point-based Value Iteration for POMDPs

TL;DR

Perseus is a randomized point-based value iteration algorithm for POMDPs that efficiently improves policy quality by backing up a subset of belief points, scalable to large problems and continuous actions.

Contribution

It introduces a novel randomized backup approach that enhances scalability and efficiency in solving large-scale POMDPs with continuous actions.

Findings

Perseus outperforms existing methods on large POMDP benchmarks.

The algorithm effectively handles continuous action spaces.

Experimental results demonstrate significant computational savings.

Abstract

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.