Creating a Reinforcement Learning Controller for Functional Electrical Stimulation of a Human Arm

Philip S Thomas; Michael Branicky; Antonie van den Bogert; Kathleen Jagodnik

Creating a Reinforcement Learning Controller for Functional Electrical Stimulation of a Human Arm

Philip S Thomas
, Michael Branicky
, Antonie van den Bogert
, Kathleen Jagodnik

Case Western Reserve University

Research output: Other contribution

Abstract

Clinical tests have shown that the dynamics of a human arm, controlled using Functional Electrical Stimulation (FES), can vary significantly between and during trials. In this paper, we study the application of Reinforcement Learning to create a controller that can adapt to these
changing dynamics of a human arm. Development and tests were done in simulation using a two-dimensional arm model and Hill-based muscle dynamics. An actor-critic architecture is used with artificial neural networks for both the actor and the critic. We begin by training it using a Proportional Derivative (PD) controller as a supervisor. We then make clinically relevant changes
to the dynamics of the arm and test the actor-critic’s ability to adapt without supervision in a reasonable number of episodes.

Original language	American English
State	Published - Jan 1 2008

Disciplines

Biomechanical Engineering

Access to Document

Creating a Reinforcement Learning Controller for Functional ElectFinal published version, 1.67 MBLicense: CC BY

Cite this

@misc{92609e55a16946aa929198eedf45ad45,

title = "Creating a Reinforcement Learning Controller for Functional Electrical Stimulation of a Human Arm",

abstract = " Clinical tests have shown that the dynamics of a human arm, controlled using Functional Electrical Stimulation (FES), can vary significantly between and during trials. In this paper, we study the application of Reinforcement Learning to create a controller that can adapt to these changing dynamics of a human arm. Development and tests were done in simulation using a two-dimensional arm model and Hill-based muscle dynamics. An actor-critic architecture is used with artificial neural networks for both the actor and the critic. We begin by training it using a Proportional Derivative (PD) controller as a supervisor. We then make clinically relevant changes to the dynamics of the arm and test the actor-critic{\textquoteright}s ability to adapt without supervision in a reasonable number of episodes.",

author = "Thomas, \{Philip S\} and Michael Branicky and \{van den Bogert\}, Antonie and Kathleen Jagodnik",

year = "2008",

month = jan,

day = "1",

language = "American English",

type = "Other",

}

TY - GEN

T1 - Creating a Reinforcement Learning Controller for Functional Electrical Stimulation of a Human Arm

AU - Thomas, Philip S

AU - Branicky, Michael

AU - van den Bogert, Antonie

AU - Jagodnik, Kathleen

PY - 2008/1/1

Y1 - 2008/1/1

N2 - Clinical tests have shown that the dynamics of a human arm, controlled using Functional Electrical Stimulation (FES), can vary significantly between and during trials. In this paper, we study the application of Reinforcement Learning to create a controller that can adapt to these changing dynamics of a human arm. Development and tests were done in simulation using a two-dimensional arm model and Hill-based muscle dynamics. An actor-critic architecture is used with artificial neural networks for both the actor and the critic. We begin by training it using a Proportional Derivative (PD) controller as a supervisor. We then make clinically relevant changes to the dynamics of the arm and test the actor-critic’s ability to adapt without supervision in a reasonable number of episodes.

AB - Clinical tests have shown that the dynamics of a human arm, controlled using Functional Electrical Stimulation (FES), can vary significantly between and during trials. In this paper, we study the application of Reinforcement Learning to create a controller that can adapt to these changing dynamics of a human arm. Development and tests were done in simulation using a two-dimensional arm model and Hill-based muscle dynamics. An actor-critic architecture is used with artificial neural networks for both the actor and the critic. We begin by training it using a Proportional Derivative (PD) controller as a supervisor. We then make clinically relevant changes to the dynamics of the arm and test the actor-critic’s ability to adapt without supervision in a reasonable number of episodes.

UR - https://engagedscholarship.csuohio.edu/enme_facpub/413

M3 - Other contribution

ER -

Creating a Reinforcement Learning Controller for Functional Electrical Stimulation of a Human Arm

Abstract

Disciplines

Access to Document

Other files and links

Cite this