Quantized measurements in Q-learning based model-free optimal control

Sini Tiistola, Risto Ritala, Matti Vilkko

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

2 Citations (Scopus)
16 Downloads (Pure)

Abstract

Quantization noise is present in many real-time applications due to the resolution of analog-to-digital conversions. This can lead to error in policies that are learned by model-free Q-learning. A method for quantization error reduction for Q-learning algorithms is developed using the sample time and an exploration noise that is added to the control input. The method is illustrated with discrete-time policy and value iteration algorithms using both a simulated environment and a real-time physical system.
Original languageEnglish
Title of host publication21th IFAC World Congress
EditorsRolf Findeisen, Sandra Hirche, Klaus Janschek, Martin Mönnigmann
PublisherElsevier
Pages1640-1645
Number of pages6
DOIs
Publication statusPublished - 2020
Publication typeA4 Article in conference proceedings
EventIFAC World Congress - Berlin, Germany
Duration: 11 Jul 202017 Jul 2020

Publication series

NameIFAC-PapersOnLine
PublisherElsevier
Number2
Volume53
ISSN (Print)2405-8971
ISSN (Electronic)2405-8963

Conference

ConferenceIFAC World Congress
Country/TerritoryGermany
CityBerlin
Period11/07/2017/07/20

Publication forum classification

  • Publication forum level 1

Fingerprint

Dive into the research topics of 'Quantized measurements in Q-learning based model-free optimal control'. Together they form a unique fingerprint.

Cite this