Development of stress-free human support robot by proactive action with semantic map

2019 Fiscal Year Final Research Report

• Project/Area Number: 17K14619
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Dynamics/Control
• Research Institution: Keio University
• Principal Investigator: Yorozu Ayanori 慶應義塾大学, 理工学研究科(矢上), 特任助教 (40781159)
• Project Period (FY): 2017-04-01 – 2020-03-31
• Keywords: セマンティックマップ / 自己位置推定 / 歩行者の姿勢推定

Outline of Final Research Achievements

For stress-free human-robot interaction, it is necessary to recognize and predict human motion based on the relationship between surrounding objects and humans. However, in the human living space, the kidnapped robot problem tends to be occurred due to moving objects. In this study, classes of objects and their movement attributes were defined and a semantic map that includes the attributes of objects were proposed. By changing the importance of the correspondence between the map and the sensor information based on the class and attribute of the object, we proposed a robust localization to prevent the problem even when the environment changes. In addition, to predict the movement of pedestrian, an estimation of body direction based on the acquired position, velocity and gait phase of both legs was proposed. Furthermore, we evaluated the psychological effect of the robot approaching human, focusing on the visual and auditory sense, and modeled a psychological spatial domain.

Free Research Field

知能ロボティクス

Academic Significance and Societal Importance of the Research Achievements

移動する歩行者や配置が変化する物体などが存在する動的な環境において、物体情報として移動や位置変化の時間スケールを考慮した属性をマップ上に付加することで、ロバストな自己位置推定を実現することが可能であり、移動ロボットの個別の技術としても汎用性が高い手法を確立した。また、足元のセンシングから上体の姿勢角を推定する技術は、ロボットに限らず、福祉分野における現場での高齢者の歩行計測などへの応用が期待される。