Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms

Vahid Tavakol Aghaei; Ahmet Onat

doi:10.1145/3068796.3068827

Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms

Vahid Tavakol Aghaei, Ahmet Onat

Sabanci University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Citation (Scopus)

Abstract

In this study a gain scheduling method for the scaling factors of the input variables to the fuzzy logic controller by means of policy gradient reinforcement learning algorithms has been proposed. The motivation for using PG algorithms is that they can scale RL problems into continuous high dimensional state-action spaces without the need for function approximation methods. Without incorporating any a-priori knowledge of the plant, the proposed method optimizes the cost function of the learning algorithm and tries to find optimal solutions for the scaling factors of the fuzzy logic controller. To show the effectiveness of the proposed method it has been applied to a PD type fuzzy controller along with a nonlinear model of an inverted pendulum. By performing different simulations, it is observed that the proposed method can find optimal solutions within a small number of learning iterations.

Original language	English
Title of host publication	Proceedings of 2017 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017
Publisher	Association for Computing Machinery
Pages	146-151
Number of pages	6
ISBN (Electronic)	9781450352802
DOIs	https://doi.org/10.1145/3068796.3068827
Publication status	Published - 8 Feb 2017
Externally published	Yes
Event	3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017 - Paris, France Duration: 8 Feb 2017 → 12 Feb 2017

Publication series

Name	ACM International Conference Proceeding Series
Volume	Part F128050

Conference

Conference	3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017
Country/Territory	France
City	Paris
Period	8/02/17 → 12/02/17

Bibliographical note

Publisher Copyright:
© 2017 Association for Computing Machinery.

Keywords

Fuzzy control
Fuzzy logic
Policy gradients
Reinforcement learning
Tuning scaling factors

Access to Document

10.1145/3068796.3068827

Cite this

Aghaei, V. T., & Onat, A. (2017). Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms. In Proceedings of 2017 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017 (pp. 146-151). (ACM International Conference Proceeding Series; Vol. Part F128050). Association for Computing Machinery. https://doi.org/10.1145/3068796.3068827

@inproceedings{8f35a8da4b54468a85cb5f677633d6a1,

title = "Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms",

abstract = "In this study a gain scheduling method for the scaling factors of the input variables to the fuzzy logic controller by means of policy gradient reinforcement learning algorithms has been proposed. The motivation for using PG algorithms is that they can scale RL problems into continuous high dimensional state-action spaces without the need for function approximation methods. Without incorporating any a-priori knowledge of the plant, the proposed method optimizes the cost function of the learning algorithm and tries to find optimal solutions for the scaling factors of the fuzzy logic controller. To show the effectiveness of the proposed method it has been applied to a PD type fuzzy controller along with a nonlinear model of an inverted pendulum. By performing different simulations, it is observed that the proposed method can find optimal solutions within a small number of learning iterations.",

keywords = "Fuzzy control, Fuzzy logic, Policy gradients, Reinforcement learning, Tuning scaling factors",

author = "Aghaei, {Vahid Tavakol} and Ahmet Onat",

note = "Publisher Copyright: {\textcopyright} 2017 Association for Computing Machinery.; 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017 ; Conference date: 08-02-2017 Through 12-02-2017",

year = "2017",

month = feb,

day = "8",

doi = "10.1145/3068796.3068827",

language = "English",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

pages = "146--151",

booktitle = "Proceedings of 2017 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017",

}

Aghaei, VT & Onat, A 2017, Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms. in Proceedings of 2017 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017. ACM International Conference Proceeding Series, vol. Part F128050, Association for Computing Machinery, pp. 146-151, 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017, Paris, France, 8/02/17. https://doi.org/10.1145/3068796.3068827

Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms. / Aghaei, Vahid Tavakol; Onat, Ahmet.
Proceedings of 2017 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017. Association for Computing Machinery, 2017. p. 146-151 (ACM International Conference Proceeding Series; Vol. Part F128050).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms

AU - Aghaei, Vahid Tavakol

AU - Onat, Ahmet

PY - 2017/2/8

Y1 - 2017/2/8

N2 - In this study a gain scheduling method for the scaling factors of the input variables to the fuzzy logic controller by means of policy gradient reinforcement learning algorithms has been proposed. The motivation for using PG algorithms is that they can scale RL problems into continuous high dimensional state-action spaces without the need for function approximation methods. Without incorporating any a-priori knowledge of the plant, the proposed method optimizes the cost function of the learning algorithm and tries to find optimal solutions for the scaling factors of the fuzzy logic controller. To show the effectiveness of the proposed method it has been applied to a PD type fuzzy controller along with a nonlinear model of an inverted pendulum. By performing different simulations, it is observed that the proposed method can find optimal solutions within a small number of learning iterations.

AB - In this study a gain scheduling method for the scaling factors of the input variables to the fuzzy logic controller by means of policy gradient reinforcement learning algorithms has been proposed. The motivation for using PG algorithms is that they can scale RL problems into continuous high dimensional state-action spaces without the need for function approximation methods. Without incorporating any a-priori knowledge of the plant, the proposed method optimizes the cost function of the learning algorithm and tries to find optimal solutions for the scaling factors of the fuzzy logic controller. To show the effectiveness of the proposed method it has been applied to a PD type fuzzy controller along with a nonlinear model of an inverted pendulum. By performing different simulations, it is observed that the proposed method can find optimal solutions within a small number of learning iterations.

KW - Fuzzy control

KW - Fuzzy logic

KW - Policy gradients

KW - Reinforcement learning

KW - Tuning scaling factors

UR - http://www.scopus.com/inward/record.url?scp=85021361879&partnerID=8YFLogxK

U2 - 10.1145/3068796.3068827

DO - 10.1145/3068796.3068827

M3 - Conference contribution

AN - SCOPUS:85021361879

T3 - ACM International Conference Proceeding Series

SP - 146

EP - 151

BT - Proceedings of 2017 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017

PB - Association for Computing Machinery

T2 - 3rd International Conference on Mechatronics and Robotics Engineering, ICMRE 2017

Y2 - 8 February 2017 through 12 February 2017

ER -

Tuning scaling factors of fuzzy logic controllers via reinforcement learning policy gradient algorithms

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this