Preliminary investigation of using deep reinforcement learning to control a mobile robot for human activity recognition

• Item type: SIG Technical Reports(1)
• 公開日: 2019-12-03
• タイトル: Preliminary investigation of using deep reinforcement learning to control a mobile robot for human activity recognition
• タイトル / 言語: en
• タイトル: Preliminary investigation of using deep reinforcement learning to control a mobile robot for human activity recognition
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 予測と認識
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University
• 著者所属: Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University
• 著者所属: Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University
• 著者所属: Araya Inc.
• 著者所属: Araya Inc.
• 著者所属(英): en
• 著者所属(英): Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University
• 著者所属(英): en
• 著者所属(英): Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University
• 著者所属(英): en
• 著者所属(英): Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University
• 著者所属(英): en
• 著者所属(英): Araya Inc.
• 著者所属(英): en
• 著者所属(英): Araya Inc.
• 著者名: Teerawat, Kumrai; Joseph, Korpela; Takuya, Maekawa; Yen, Yu; Ryota, Kanai
• 著者名(英): Teerawat, Kumrai; Joseph, Korpela; Takuya, Maekawa; Yen, Yu; Ryota, Kanai
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Due to recent advances in robotics technologies, it is becoming feasible for mobile robots to use their sensors to observe daily human activities for the purpose of human activity recognition (HAR) in indoor environments. However, when doing so, the robot will have difficultly observing and recognizing human activities when it is positioned behind the human or some obstacle. Therefore, this work investigates a method for using deep reinforcement learning to control the mobile robot's movement when observing human activities. Our objective is to minimize the movement of the robot (i.e., its energy consumption) while maximizing its human activity recognition accuracy. Moreover, our method introduces a new HAR method based on skeletal and visual features extracted from the robot's captured images.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Due to recent advances in robotics technologies, it is becoming feasible for mobile robots to use their sensors to observe daily human activities for the purpose of human activity recognition (HAR) in indoor environments. However, when doing so, the robot will have difficultly observing and recognizing human activities when it is positioned behind the human or some obstacle. Therefore, this work investigates a method for using deep reinforcement learning to control the mobile robot's movement when observing human activities. Our objective is to minimize the movement of the robot (i.e., its energy consumption) while maximizing its human activity recognition accuracy. Moreover, our method introduces a new HAR method based on skeletal and visual features extracted from the robot's captured images.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA1221543X
• 書誌情報: 研究報告ヒューマンコンピュータインタラクション（HCI）; 巻 2019-HCI-185, 号 12, p. 1-8, 発行日 2019-12-03
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8760
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会