Strong probabilistic planning

Silvio Do Lago Pereira, Leliane Nunes De Barros, Fábio Gagliardi Cozman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Scopus citations

Abstract

We consider the problem of synthesizing policies, in domains where actions have probabilistic effects, that are optimal in the expected-case among the optimal worst-case strong policies. Thus we combine features from nondeterministic and probabilistic planning in a single framework. We present an algorithm that combines dynamic programming and model checking techniques to find plans satisfying the problem requirements: the strong preimage computation from model checking is used to avoid actions that lead to cycles or dead ends, reducing the model to a Markov Decision Process where all possible policies are strong and worst-case optimal (i.e., successful and minimum length with probability 1). We show that backward induction can then be used to select a policy in this reduced model. The resulting algorithm is presented in two versions (enumerative and symbolic); we show that the latter version allows planning with extended reachability goals. © 2008 Springer Berlin Heidelberg.
Original languageAmerican English
Title of host publicationStrong probabilistic planning
Pages636-652
Number of pages17
ISBN (Electronic)3540886354, 9783540886358
DOIs
StatePublished - 5 Dec 2008
Externally publishedYes
EventLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) -
Duration: 1 Jan 2018 → …

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5317 LNAI
ISSN (Print)0302-9743

Conference

ConferenceLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Period1/01/18 → …

Fingerprint

Dive into the research topics of 'Strong probabilistic planning'. Together they form a unique fingerprint.

Cite this