Information-seeking polynomial NARX model-predictive control through expected free energy minimization

Wouter M. Kouw

doi:10.1109/LCSYS.2023.3347190

Information-seeking polynomial NARX model-predictive control through expected free energy minimization

Wouter M. Kouw

Research output: Contribution to journal › Article › Academic › peer-review

1 Citation (Scopus)

19 Downloads (Pure)

Abstract

We propose an adaptive model-predictive controller that balances driving the system to a goal state and seeking system observations that are informative with respect to the parameters of a nonlinear autoregressive exogenous model. The controller's objective function is derived from an expected free energy functional and contains information-theoretic terms expressing uncertainty over model parameters and output predictions. Experiments demonstrate that the proposed controller dynamically balances information-seeking and goal-seeking behaviour based on parameter uncertainty.

Original language	English
Article number	10373893
Pages (from-to)	37-42
Number of pages	6
Journal	IEEE Control Systems Letters
Volume	8
Early online date	25 Dec 2023
DOIs	https://doi.org/10.1109/LCSYS.2023.3347190
Publication status	Published - 2024

Keywords

Bayesian inference
Free energy minimization
Adaptive control
System identification
Adaptation models
Uncertainty
Estimation
Predictive models
Minimization
free energy minimization
Bayes methods
system identification
Information theory

Access to Document

10.1109/LCSYS.2023.3347190

pdfFinal published version, 713 KBLicence: TAVERNE

https://arxiv.org/pdf/2312.15046.pdf

Cite this

@article{3eb669930eda4907bd2d75230c1b28dd,

title = "Information-seeking polynomial NARX model-predictive control through expected free energy minimization",

abstract = "We propose an adaptive model-predictive controller that balances driving the system to a goal state and seeking system observations that are informative with respect to the parameters of a nonlinear autoregressive exogenous model. The controller's objective function is derived from an expected free energy functional and contains information-theoretic terms expressing uncertainty over model parameters and output predictions. Experiments demonstrate that the proposed controller dynamically balances information-seeking and goal-seeking behaviour based on parameter uncertainty.",

keywords = "Bayesian inference, Free energy minimization, Adaptive control, System identification, Adaptation models, Uncertainty, Estimation, Predictive models, Minimization, free energy minimization, Bayes methods, system identification, Information theory",

author = "Kouw, {Wouter M.}",

year = "2024",

doi = "10.1109/LCSYS.2023.3347190",

language = "English",

volume = "8",

pages = "37--42",

journal = "IEEE Control Systems Letters",

issn = "2475-1456",

publisher = "Institute of Electrical and Electronics Engineers",

}

TY - JOUR

T1 - Information-seeking polynomial NARX model-predictive control through expected free energy minimization

AU - Kouw, Wouter M.

PY - 2024

Y1 - 2024

N2 - We propose an adaptive model-predictive controller that balances driving the system to a goal state and seeking system observations that are informative with respect to the parameters of a nonlinear autoregressive exogenous model. The controller's objective function is derived from an expected free energy functional and contains information-theoretic terms expressing uncertainty over model parameters and output predictions. Experiments demonstrate that the proposed controller dynamically balances information-seeking and goal-seeking behaviour based on parameter uncertainty.

AB - We propose an adaptive model-predictive controller that balances driving the system to a goal state and seeking system observations that are informative with respect to the parameters of a nonlinear autoregressive exogenous model. The controller's objective function is derived from an expected free energy functional and contains information-theoretic terms expressing uncertainty over model parameters and output predictions. Experiments demonstrate that the proposed controller dynamically balances information-seeking and goal-seeking behaviour based on parameter uncertainty.

KW - Bayesian inference

KW - Free energy minimization

KW - Adaptive control

KW - System identification

KW - Adaptation models

KW - Uncertainty

KW - Estimation

KW - Predictive models

KW - Minimization

KW - free energy minimization

KW - Bayes methods

KW - system identification

KW - Information theory

UR - http://www.scopus.com/inward/record.url?scp=85181578039&partnerID=8YFLogxK

U2 - 10.1109/LCSYS.2023.3347190

DO - 10.1109/LCSYS.2023.3347190

M3 - Article

SN - 2475-1456

VL - 8

SP - 37

EP - 42

JO - IEEE Control Systems Letters

JF - IEEE Control Systems Letters

M1 - 10373893

ER -

Information-seeking polynomial NARX model-predictive control through expected free energy minimization

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

CRT-STA-FEP-walker LWAI: Learning to Walk by Active Inference

Cite this

Information-seeking polynomial NARX model-predictive control through expected free energy minimization

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Projects

CRT-STA-FEP-walker LWAI: Learning to Walk by Active Inference

Cite this