Online Algorithms for POMDPs with Continuous State, Action, and Observation Spaces
Proceedings of the International Conference on Automated Planning and Scheduling · 15.06.2018 · 28 · 259-263
DOI: 10.1609/icaps.v28i1.13882 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the International Conference on Automated Planning and Scheduling, online erschienen am 15.06.2018. Verfasst von Zachary Sunberg und 1 weiteren Autor:innen. Die Arbeit wurde bisher 98-mal zitiert (Crossref).
Abstract (Originalsprache)
Online solvers for partially observable Markov decision processes have been applied to problems with large discrete state spaces, but continuous state, action, and observation spaces remain a challenge. This paper begins by investigating double progressive widening (DPW) as a solution to this challenge. However, we prove that this modification alone is not sufficient because the belief representations in the search tree collapse to a single particle causing the algorithm to converge to a policy that is suboptimal regardless of the computation time. This paper proposes and evaluates two new algorithms, POMCPOW and PFT-DPW, that overcome this deficiency by using weighted particle filtering. Simulation results show that these modifications allow the algorithms to be successful where previous approaches fail.
Zitieren
APA
Sunberg, Z., & Kochenderfer, M. (2018). Online Algorithms for POMDPs with Continuous State, Action, and Observation Spaces. Proceedings of the International Conference on Automated Planning and Scheduling, 28, 259-263. https://doi.org/10.1609/icaps.v28i1.13882
Vancouver
Sunberg Z, Kochenderfer M. Online Algorithms for POMDPs with Continuous State, Action, and Observation Spaces. Proceedings of the International Conference on Automated Planning and Scheduling. 2018;28:259-263. doi:10.1609/icaps.v28i1.13882
BibTeX
@article{Sunberg2018,
author = {Sunberg, Zachary and Kochenderfer, Mykel},
title = {Online Algorithms for POMDPs with Continuous State, Action, and Observation Spaces},
journal = {Proceedings of the International Conference on Automated Planning and Scheduling},
year = {2018},
volume = {28},
pages = {259--263},
doi = {10.1609/icaps.v28i1.13882}
}
EUVIMED