Building a Theoretical Framework of Bodily Action: Segmentation, Recognition, and Generation of Actions Based on Dimensional Invariants.

2019 Fiscal Year Final Research Report

• Project/Area Number: 16H05860
• Research Category: Grant-in-Aid for Young Scientists (A)
• Allocation Type: Single-year Grants
• Research Field: Cognitive science
• Research Institution: Japan Advanced Institute of Science and Technology
• Principal Investigator: Hidaka Shohei 北陸先端科学技術大学院大学, 先端科学技術研究科, 准教授 (50582912)
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: 模倣 / 運動 / 身体 / 力学系 / 意図推定

Outline of Final Research Achievements

Many of our cognitive activities are performed through physical interaction with others. It has been considered that bodily imitation plays one of the central roles in such a social interaction. It is necessary to convert visual representation of other’s bodily action to one’s own motor command, if one performs a bodily imitation, and this requires cross-modal mapping among multiple sensors and effectors. In this study, we built a theoretical foundation of recognition and segmentation of action, and demonstrated it with a time series of bodily movement a in an unsupervised manner based on invariant feature of an underlying dynamical system.

Free Research Field

認知科学

Academic Significance and Societal Importance of the Research Achievements

身体模倣に内在する基本的な性質を解明することで、我々が他者の動きからどのような情報を得ているか、相手がどんな意図を持つか、などの基礎的なメカニズムの理解へとつながると期待される。将来的には、具体的に、自閉症などの社会的な認知機能の障害とみられる傾向をもつ方の支援や周囲の人たちがとるべきコミュニケーションの方法についての示唆を与えることにつながると考えられる。