On Speeding Up Methods for Identifying Redundant Actions in Plans

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F22%3A00363584" target="_blank" >RIV/68407700:21230/22:00363584 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1609/icaps.v32i1.19808" target="_blank" >https://doi.org/10.1609/icaps.v32i1.19808</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1609/icaps.v32i1.19808" target="_blank" >10.1609/icaps.v32i1.19808</a>

Jazyk výsledku

angličtina

Název v původním jazyce

On Speeding Up Methods for Identifying Redundant Actions in Plans

Popis výsledku v původním jazyce

Satisficing planning aims at generating plans that are not necessarily optimal. Often, minimising plan generation time negatively affects quality of generated plans. Acquiring plans quickly might be of critical importance in decision-making systems that operate nearly in realtime. However, (very) suboptimal plans might be expensive to execute and more prone to failures. Optimising plans after they are generated, in a spare time, can improve their quality. This paper focuses on speeding up the (Greedy) Action Elimination methods, which are used for identifying and removing redundant actions from plans in polynomial time. We present two enhancements of these methods: Plan Action Landmarks, actions that are not redundant in a given plan, and Action Cycles which are subsequences of actions which if removed do not affect the state trajectory after the last action of the cycle. We evaluate the introduced methods on benchmark problems from the Agile tracks of the International Planning Competition and on plans generated by several state-of-the-art planners, successful in the recent editions of the competition. 2022, Association for the Advancement of Artificial Intelligence.

Název v anglickém jazyce

On Speeding Up Methods for Identifying Redundant Actions in Plans

Popis výsledku anglicky

Satisficing planning aims at generating plans that are not necessarily optimal. Often, minimising plan generation time negatively affects quality of generated plans. Acquiring plans quickly might be of critical importance in decision-making systems that operate nearly in realtime. However, (very) suboptimal plans might be expensive to execute and more prone to failures. Optimising plans after they are generated, in a spare time, can improve their quality. This paper focuses on speeding up the (Greedy) Action Elimination methods, which are used for identifying and removing redundant actions from plans in polynomial time. We present two enhancements of these methods: Plan Action Landmarks, actions that are not redundant in a given plan, and Action Cycles which are subsequences of actions which if removed do not affect the state trajectory after the last action of the cycle. We evaluate the introduced methods on benchmark problems from the Agile tracks of the International Planning Competition and on plans generated by several state-of-the-art planners, successful in the recent editions of the competition. 2022, Association for the Advancement of Artificial Intelligence.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

Výsledek vznikl pri realizaci vícero projektů. Více informací v záložce Projekty.

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of the Thirty-Second International Conference on Automated Planning and Scheduling, ICAPS 2022

ISBN

978-1-57735-874-9

ISSN

2334-0835

e-ISSN

—

Počet stran výsledku

9

Strana od-do

252-260

Název nakladatele

AAAI Press

Místo vydání

Menlo Park

Místo konání akce

Singapur

Datum konání akce

19. 6. 2022

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—