Planning and Acting with Non-deterministic Events:Navigating between Safe States

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F20%3A00341720" target="_blank" >RIV/68407700:21230/20:00341720 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.aaai.org/Papers/AAAI/2020GB/AAAI-ChrpaL.7590.pdf" target="_blank" >https://www.aaai.org/Papers/AAAI/2020GB/AAAI-ChrpaL.7590.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Planning and Acting with Non-deterministic Events:Navigating between Safe States

Popis výsledku v původním jazyce

Automated Planning addresses the problem of finding a se-quence of actions, a plan, transforming the environment fromits initial state to some goal state. In real-world environments,exogenous events might occur and might modify the envi-ronment without agent’s consent. Besides disrupting agent’splan, events might hinder agent’s pursuit towards its goals andeven cause damage (e.g. destroying the robot).In this paper, we leverage the notion ofSafe Statesin dynamicenvironments under presence of non-deterministic exogenousevents that might eventually cause dead-ends (e.g. “damage”the agent) if the agent is not careful while executing its plan.We introduce a technique for generating plans that constrainsthe number of consecutive “unsafe” actions in a plan and atechnique for generating “robust” plans that effectively evadeevent effects. Combination of both approaches plans and ex-ecutes robust plans between safe states. We empirically showthat such an approach effectively navigates the agent towardsits goals in spite of presence of dead-ends.

Název v anglickém jazyce

Planning and Acting with Non-deterministic Events:Navigating between Safe States

Popis výsledku anglicky

Automated Planning addresses the problem of finding a se-quence of actions, a plan, transforming the environment fromits initial state to some goal state. In real-world environments,exogenous events might occur and might modify the envi-ronment without agent’s consent. Besides disrupting agent’splan, events might hinder agent’s pursuit towards its goals andeven cause damage (e.g. destroying the robot).In this paper, we leverage the notion ofSafe Statesin dynamicenvironments under presence of non-deterministic exogenousevents that might eventually cause dead-ends (e.g. “damage”the agent) if the agent is not careful while executing its plan.We introduce a technique for generating plans that constrainsthe number of consecutive “unsafe” actions in a plan and atechnique for generating “robust” plans that effectively evadeevent effects. Combination of both approaches plans and ex-ecutes robust plans between safe states. We empirically showthat such an approach effectively navigates the agent towardsits goals in spite of presence of dead-ends.

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

<a href="/cs/project/EF16_019%2F0000765" target="_blank" >EF16_019/0000765: Výzkumné centrum informatiky</a><br>

Návaznosti

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

Rok uplatnění

2020

Kód důvěrnosti údajů

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

Název statě ve sborníku

Proceedings of the AAAI Conference on Artificial Intelligence

ISBN

—

ISSN

2159-5399

e-ISSN

2374-3468

Počet stran výsledku

8

Strana od-do

9802-9809

Název nakladatele

AAAI Press

Místo vydání

Menlo Park

Místo konání akce

New York

Datum konání akce

7. 2. 2020

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—