About
Contact
Help
Sending publications
How to publish
Advanced Search
View Item 
  •   Home
  • Facultad de Ciencias Físicas y Matemáticas
  • Artículos de revistas
  • View Item
  •   Home
  • Facultad de Ciencias Físicas y Matemáticas
  • Artículos de revistas
  • View Item
JavaScript is disabled for your browser. Some features of this site may not work without it.

Browse byCommunities and CollectionsDateAuthorsTitlesSubjectsThis CollectionDateAuthorsTitlesSubjects

My Account

Login to my accountRegister
Biblioteca Digital - Universidad de Chile
Revistas Chilenas
Repositorios Latinoamericanos
Tesis LatinoAmericanas
Tesis chilenas
Related linksRegistry of Open Access RepositoriesOpenDOARGoogle scholarCOREBASE
My Account
Login to my accountRegister

Toward real-time decentralized reinforcement learning using finite support basis functions

Artículo
Thumbnail
Open/Download
Iconitem_85053882225.pdf (2.037Kb)
Access note
Acceso a solo metadatos
Publication date
2017
Metadata
Show full item record
Cómo citar
Lobos-Tsunekawa, Kenzo
Cómo citar
Toward real-time decentralized reinforcement learning using finite support basis functions
.
Copiar
Cerrar

Author
  • Lobos-Tsunekawa, Kenzo;
  • Leottau, David;
  • Ruiz del Solar, Javier;
Abstract
This paper addresses the design and implementation of complex Reinforcement Learning (RL) behaviors where multi-dimensional action spaces are involved, as well as the need to execute the behaviors in real-time using robotic platforms with limited computational resources and training times. For this purpose, we propose the use of decentralized RL, in combination with finite support basis functions as alternatives to Gaussian RBF, in order to alleviate the effects of the curse of dimensionality on the action and state spaces respectively, and to reduce the computation time. As testbed, a RL based controller for the in-walk kick in NAO robots, a challenging and critical problem for soccer robotics, is used. The reported experiments show empirically that our solution saves up to 99.94% of execution time and 98.82% of memory consumption during execution, without diminishing performance compared to classical approaches.
Indexation
Artículo de publicación SCOPUS
Identifier
URI: https://repositorio.uchile.cl/handle/2250/169506
DOI: 10.1007/978-3-030-00308-1_8
ISSN: 16113349
03029743
Quote Item
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Volumen 11175 LNAI, 2017
Collections
  • Artículos de revistas
xmlui.footer.title
31 participating institutions
More than 73,000 publications
More than 110,000 topics
More than 75,000 authors
Published in the repository
  • How to publish
  • Definitions
  • Copyright
  • Frequent questions
Documents
  • Dating Guide
  • Thesis authorization
  • Document authorization
  • How to prepare a thesis (PDF)
Services
  • Digital library
  • Chilean academic journals portal
  • Latin American Repository Network
  • Latin American theses
  • Chilean theses
Dirección de Servicios de Información y Bibliotecas (SISIB)
Universidad de Chile

© 2020 DSpace
  • Access my account