2A1-E07 CTRNNを用いた連続な状態空間における強化学習法の提案

• 有江 浩明

  早大

• 尾形 哲也

  京大

• 谷 淳

  理研

• 菅野 重樹

  早大

書誌事項

タイトル別名

• 2A1-E07 Reinforcement Learning with CTRNN in Continuous Space

抄録

There are some difficulties to apply traditional reinforcement learning algorithms to robot motion control task. Because most algorithms are concerned with discrete state space and based on the assumption of complete observability of the state. This paper deals with these problems by combining the reinforcement learning algorithm and CTRNN learning algorithm (BPTT). We carried out an experiment on pendulum swing up task without rotational speed information. It is shown that the information about rotational speed of pendulum, which is considered as a hidden state, is estimated and encoded on the activation of a context unit.

収録刊行物

• ロボティクス・メカトロニクス講演会講演概要集

  ロボティクス・メカトロニクス講演会講演概要集 2006 (0), _2A1-E07_1-_2A1-E07_2, 2006

  一般社団法人 日本機械学会

キーワード

• Reinforcement Learning
• Neural Network

詳細情報

• CRID: 1390282680913177984
• NII論文ID: 110008694061
• DOI: 10.1299/jsmermd.2006._2a1-e07_1
• ISSN: 24243124
• Web Site: https://www.jstage.jst.go.jp/article/jsmermd/2006/0/2006__2A1-E07_1/_pdf
• 本文言語コード: ja
• データソース種別:
  • JaLC
  • Crossref
  • CiNii Articles
• 抄録ライセンスフラグ: 使用不可