筋骨格システムを対象にした筋内力フィードフォワード位置制御法における強化学習を用いた筋内力決定法

Translated title of the contribution: Muscular internal force determination method using reinforcement learning for feedforward positioning of musculoskeletal system

松谷 祐希, 田原 健二, 木野 仁, 越智 裕章, 山本 元司

Research output: Contribution to journalArticlepeer-review

Abstract

The purpose of this paper is to present solution to an ill-posed problem for a muscular internal force feedforward positioning method of a musculoskeletal system. In our previous research, the muscular internal force feedforward positioning method of the musculoskeletal system has been proposed. In the method, the position regulation of the system can be accomplished by inputting a desired internal force balancing at a desired position. However, this control method has the ill-posed problem that the muscular internal force balancing at the desired position cannot be uniquely determined because the musculoskeletal system has muscular redundancy. A determination method of the muscular internal force is an important problem because a convergence and responsiveness of the system are influenced by the muscular internal force. Therefore, this study proposes a new determination method of the muscular internal force using reinforcement learning technique in order to determine the muscular internal force uniquely by considering control performance of the system. The proposed method numerically determines the muscular internal force that can converge at a desired position smaller than a conventional method. Its effectiveness is shown through numerical simulations for reaching movements of the musculoskeletal system.
Translated title of the contributionMuscular internal force determination method using reinforcement learning for feedforward positioning of musculoskeletal system
Original languageJapanese
Pages (from-to)14-00313-14-00313
Journal日本機械学会論文集
Volume81
Issue number822
DOIs
Publication statusPublished - 2015

Fingerprint Dive into the research topics of 'Muscular internal force determination method using reinforcement learning for feedforward positioning of musculoskeletal system'. Together they form a unique fingerprint.

Cite this