Improving the Performance of Backward Chained Behavior Trees that use Reinforcement Learning

↓ Direkt till sidans innehåll
↓ Direkt till sidans sekundära innehåll (sidomenyn)

Sökning: onr:"swepub:oai:DiVA.org:kth-342643" > Improving the Perfo...

1 av 1
Föregående post
Nästa post
Till träfflistan

LIBRIS Formathandbok (Information om MARC21)

Fältnamn	Indikatorer	Metadata
000		03522naa a2200385 4500
001		oai:DiVA.org:kth-342643
003		SwePub
008		240125s2023 \| \|\|\|\|\|\|\|\|\|\|\|000 \|\|eng\|
024	7	a https://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-3426432 URI
024	7	a https://doi.org/10.1109/IROS55552.2023.103423192 DOI
040		a (SwePub)kth
041		a engb eng
042		9 SwePub
072	7	a ref2 swepub-contenttype
072	7	a kon2 swepub-publicationtype
100	1	a Kartasev, Martu KTH,Robotik, perception och lärande, RPL4 aut0 (Swepub:kth)u10b731a
245	1 0	a Improving the Performance of Backward Chained Behavior Trees that use Reinforcement Learning
264	1	b Institute of Electrical and Electronics Engineers (IEEE),c 2023
338		a print2 rdacarrier
500		a Part of ISBN 978-1-6654-9190-7QC 20240130
520		a In this paper we show how to improve the performance of backward chained behavior trees (BTs) that include policies trained with reinforcement learning (RL). BTs represent a hierarchical and modular way of combining control policies into higher level control policies. Backward chaining is a design principle for the construction of BTs that combines reactivity with goal directed actions in a structured way. The backward chained structure has also enabled convergence proofs for BTs, identifying a set of local conditions to be satisfied for the convergence of all trajectories to a set of desired goal states. The key idea of this paper is to improve performance of backward chained BTs by using the conditions identified in a theoretical convergence proof to configure the RL problems for individual controllers. Specifically, previous analysis identified so-called active constraint conditions (ACCs), that should not be violated in order to avoid having to return to work on previously achieved subgoals. We propose a way to set up the RL problems, such that they do not only achieve each immediate subgoal, but also avoid violating the identified ACCs. The resulting performance improvement depends on how often ACC violations occurred before the change, and how much effort, in terms of execution time, was needed to re-achieve them. The proposed approach is illustrated in a dynamic simulation environment.
650	7	a TEKNIK OCH TEKNOLOGIERx Elektroteknik och elektronikx Robotteknik och automation0 (SwePub)202012 hsv//swe
650	7	a ENGINEERING AND TECHNOLOGYx Electrical Engineering, Electronic Engineering, Information Engineeringx Robotics0 (SwePub)202012 hsv//eng
650	7	a NATURVETENSKAPx Data- och informationsvetenskapx Datavetenskap0 (SwePub)102012 hsv//swe
650	7	a NATURAL SCIENCESx Computer and Information Sciencesx Computer Sciences0 (SwePub)102012 hsv//eng
653		a Artificial Intelligence
653		a Autonomous systems
653		a Behavior trees
653		a Reinforcement learning
700	1	a Salér, Justinu KTH,Robotik, perception och lärande, RPL4 aut0 (Swepub:kth)u1puhfk3
700	1	a Ögren, Petter,d 1974-u KTH,Optimeringslära och systemteori,Robotik, perception och lärande, RPL4 aut0 (Swepub:kth)u1izkr9z
710	2	a KTHb Robotik, perception och lärande, RPL4 org
773	0	t 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023d : Institute of Electrical and Electronics Engineers (IEEE)g , s. 1572-1579q <1572-1579
856	4 8	u https://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-342643
856	4 8	u https://doi.org/10.1109/IROS55552.2023.10342319

Till lärosätets databas

1 av 1
Föregående post
Nästa post
Till träfflistan

Hitta mer i SwePub

Av författaren/redakt...: Kartasev, Mart; Salér, Justin; Ögren, Petter, 1 ...

Om ämnet

TEKNIK OCH TEKNOLOGIER: TEKNIK OCH TEKNO ...; och Elektroteknik oc ...; och Robotteknik och ...

NATURVETENSKAP: NATURVETENSKAP; och Data och informa ...; och Datavetenskap

Artiklar i publikationen: 2023 IEEE/RSJ In ...

Av lärosätet: Kungliga Tekniska Högskolan

Sök utanför SwePub

Sök vidare i:: Google; Google Book Search; Google Scholar

Kungliga biblioteket hanterar dina personuppgifter i enlighet med EU:s dataskyddsförordning (2018), GDPR. Läs mer om hur det funkar här.
Så här hanterar KB dina uppgifter vid användning av denna tjänst.

LIBRIS.kb.se