Motor Learning System Compensating the Characteristics of Pneumatic Manipulators

DOI Web Site Web Site 7 References
  • IKEMOTO Shuhei
    Japan Science and Technology Agency, ERATO, Asada Synergistic Intelligence Project Graduate School of Engineering, Osaka University
  • MINATO Takashi
    Japan Science and Technology Agency, ERATO, Asada Synergistic Intelligence Project
  • ISHIGURO Hiroshi
    Japan Science and Technology Agency, ERATO, Asada Synergistic Intelligence Project Graduate School of Engineering, Osaka University

Bibliographic Information

Other Title
  • 空気圧マニピュレータの特性を考慮した運動学習システムの提案
  • クウキアツ マニピュレータ ノ トクセイ オ コウリョシタ ウンドウ ガクシュウ システム ノ テイアン

Search this article

Abstract

Pneumatic actuators are suitable for humanoid robots interacting with persons in terms of size and safety since they have high flexibility and high power to weight ratio. However, it is known that it is difficult to control them stably by a feedback control because of their large dead time and high-order delay. Therefore, when we apply the feedback error learning to learn an inverse model of a system includes such actuators, the strong non-linearity makes the learning unstable. This paper focuses on an android which adopts pneumatic actuators with strong non-linearity and proposes a method to learn the inverse model of the android by compensating the characteristics of the actuators.

Journal

References(7)*help

See more

Related Projects

See more

Keywords

Details 詳細情報について

Report a problem

Back to top