Evolutionary Reinforcement Learning: Hybrid Approach for Safety-Informed Fault-Tolerant Flight Control

Vlad Gavra; Erik Jan van Kampen

doi:10.2514/1.G008112

Evolutionary Reinforcement Learning: Hybrid Approach for Safety-Informed Fault-Tolerant Flight Control

Vlad Gavra^*, Erik Jan van Kampen

^*Corresponding author for this work

Control & Simulation

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

3 Downloads (Pure)

Abstract

Recent research in artificial intelligence potentially provides solutions to the challenging problem of fault-tolerant and robust flight control. This paper proposes a novel Safety-Informed Evolutionary Reinforcement Learning algorithm (SERL), which combines Deep Reinforcement Learning (DRL) and neuroevolution to optimize a population of nonlinear control policies. Using SERL, the work has trained agents to provide attitude tracking on a high-fidelity nonlinear fixed-wing aircraft model. Compared to a state-of-the-art DRL solution, SERL achieves better tracking performance in nine out of ten cases, remaining robust against faults and changes in flight conditions, while providing smoother action signals.

Original language	English
Pages (from-to)	887-900
Number of pages	14
Journal	Journal of Guidance, Control, and Dynamics
Volume	47
Issue number	5
DOIs	https://doi.org/10.2514/1.G008112
Publication status	Published - 2024

Bibliographical note

Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Access to Document

10.2514/1.G008112

gavra-van-kampen-2024-evolutionary-reinforcement-learning-hybrid-approach-for-safety-informed-fault-tolerant-flightFinal published version, 5.89 MB

Cite this

@article{e1fa22ed84a14f6c9895abe1146f89b6,

title = "Evolutionary Reinforcement Learning: Hybrid Approach for Safety-Informed Fault-Tolerant Flight Control",

abstract = "Recent research in artificial intelligence potentially provides solutions to the challenging problem of fault-tolerant and robust flight control. This paper proposes a novel Safety-Informed Evolutionary Reinforcement Learning algorithm (SERL), which combines Deep Reinforcement Learning (DRL) and neuroevolution to optimize a population of nonlinear control policies. Using SERL, the work has trained agents to provide attitude tracking on a high-fidelity nonlinear fixed-wing aircraft model. Compared to a state-of-the-art DRL solution, SERL achieves better tracking performance in nine out of ten cases, remaining robust against faults and changes in flight conditions, while providing smoother action signals.",

author = "Vlad Gavra and {van Kampen}, {Erik Jan}",

note = "Green Open Access added to TU Delft Institutional Repository {\textquoteleft}You share, we take care!{\textquoteright} – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public. ",

year = "2024",

doi = "10.2514/1.G008112",

language = "English",

volume = "47",

pages = "887--900",

journal = "Journal of Guidance, Control, and Dynamics",

issn = "0731-5090",

publisher = "American Institute of Aeronautics and Astronautics Inc. (AIAA)",

number = "5",

}

TY - JOUR

T1 - Evolutionary Reinforcement Learning

T2 - Hybrid Approach for Safety-Informed Fault-Tolerant Flight Control

AU - Gavra, Vlad

AU - van Kampen, Erik Jan

N1 - Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2024

Y1 - 2024

N2 - Recent research in artificial intelligence potentially provides solutions to the challenging problem of fault-tolerant and robust flight control. This paper proposes a novel Safety-Informed Evolutionary Reinforcement Learning algorithm (SERL), which combines Deep Reinforcement Learning (DRL) and neuroevolution to optimize a population of nonlinear control policies. Using SERL, the work has trained agents to provide attitude tracking on a high-fidelity nonlinear fixed-wing aircraft model. Compared to a state-of-the-art DRL solution, SERL achieves better tracking performance in nine out of ten cases, remaining robust against faults and changes in flight conditions, while providing smoother action signals.

AB - Recent research in artificial intelligence potentially provides solutions to the challenging problem of fault-tolerant and robust flight control. This paper proposes a novel Safety-Informed Evolutionary Reinforcement Learning algorithm (SERL), which combines Deep Reinforcement Learning (DRL) and neuroevolution to optimize a population of nonlinear control policies. Using SERL, the work has trained agents to provide attitude tracking on a high-fidelity nonlinear fixed-wing aircraft model. Compared to a state-of-the-art DRL solution, SERL achieves better tracking performance in nine out of ten cases, remaining robust against faults and changes in flight conditions, while providing smoother action signals.

UR - http://www.scopus.com/inward/record.url?scp=85192173658&partnerID=8YFLogxK

U2 - 10.2514/1.G008112

DO - 10.2514/1.G008112

M3 - Article

AN - SCOPUS:85192173658

SN - 0731-5090

VL - 47

SP - 887

EP - 900

JO - Journal of Guidance, Control, and Dynamics

JF - Journal of Guidance, Control, and Dynamics

IS - 5

ER -

Evolutionary Reinforcement Learning: Hybrid Approach for Safety-Informed Fault-Tolerant Flight Control

Abstract

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this