Adaptive motion control of social robot for guiding a users'group under dynamic environment

• 資料種別: 博士論文
• タイトル: Adaptive motion control of social robot for guiding a users'group under dynamic environment
• 著者・編者: 張, 斌; Bin, Zhang
• 著者標目: 張, 斌; Bin, Zhang
• 出版年月日等: 2017-03-24
• 出版年（W3CDTF）: 2017-03-24
• 並列タイトル等: 動的環境下でユーザグループの案内を行うソーシャルロボットにおける適応移動制御
• 授与機関名: 電気通信大学
• 授与年月日: 2017-03-24
• 授与年月日（W3CDTF）: 2017-03-24
• 報告番号: 甲第891号
• 学位: 博士（工学）
• 博論授与番号: 甲第891号
• 本文の言語コード: eng
• 対象利用者: 一般
• 一般注記: In recent years, human-robot interaction has become one of the fastest growing research fields. One important function of service robots used in different fields is to guide users from one place to another. Conventional researches about guide robot focus on realizing the guide task and adding the functions of the guide robot by changing the motion states of the robot case by case. The guide robot requires the users to follow it or maintain a fixed distance from it. In this thesis, the purpose is to make a social guide robot that can provide adaptive social guide service to the users. Social guide robot means that the robot needs to adapt to users' activities and always accompany with them during the guiding process instead of limiting the users' motions. To realize this purpose, the environment understanding function, users group tracking function and adaptive motion control model are added to the guide robot. This thesis is organized as follows. Chapter 1 is the introduction part. It introduced the background and the conventional researches about autonomous guide robots first. After explaining the problems and tasks about the conventional researches, the purpose of this thesis is introduced. Chapter 2 explained, in detail, the path planning method for the guide robot. A 2D map of the work environment is generated in advance by Simultaneous Localization and Mapping (SLAM) method, which is usually used to generate a 2D map. The 2D map is generated by combining 3D information from Kinect sensor and 2D information from LRF sensor on the robot. With 3D information from Kinect sensor, humans and objects like walls, desks or chairs can be recognized. By reecting the immobile areas like walls on the map quickly and deleting the mobile areas like humans from the map, a more robust map for path planning is generated. The path planning part is based on combing the global and local path planning methods. Dijkstra's Algorithm is used for global path planning and the shortest path is generated and updated online. Potential field method is used for local path planning by moving towards the sub-goals, which are sampled from the shortest path. In this way, the guide robot can deal with dynamic environment and guide the users in the shortest way. Chapter 3 explained the way to recognize and track the particular users' group when the robot conducts the guide task. The problem is solved by using MCMC particle filter. Each of the users is detected and recognized by the face and color information. The recognition result is used as the observation likelihood in the tracking process. Each of the users is tracked by an independent group of particles and the tracking results are modified to the nearest human area. All of the human areas are allowed to be shared by multiple users since people may be closed with each other and form big areas containing multiple users. The effectiveness of this method was proved through tracking different groups of users under lab environment, and furthermore it is successfully applied to continuous monitoring of multiple children in a nursery school. Chapter 4 introduced the proposed framework to control the robot adaptively. In this framework, special artificial potential fields for the users and the sub-goals are generated separately, and integrated with that generated from the obstacles in the environment. The robot is controlled by the integrated potential _led and moving towards the point with the lowest potential. In this way, the guide robot can adapt to the users' activities and provide sociable tour guide services. The robot predicts the moving speed of the users' group and adapts to it to maintain the social distance. Moreover, with the proposed framework, users can deviate from the guided path temporarily and return to the original task afterward. Instead of waiting for the users andtaking the risk of losing them, the robot deviates from its original path to follow the users while preparing for returning to the guiding task. The robot restarts the guiding task from the best posture that ensures the mode changing process smoothly. The adaptive motions can be automatically generated by the proposed framework. The guide robot controlled by the proposed framework can always accompany with the users during the guide tasks and provide them guide services. The effectiveness of the proposed framework was demonstrated by simulations, and the users were guided to their destination in a sociable way by the robot in the experiments, in which different groups of users with different numbers were guided by the robot. Chapter 5 described the conclusion of this research. In this thesis, a social guide robot is designed by adapting the motions of the users group and accompanying them during the guide task. Different from the conventional works that were focused on realizing the guide task, this work tried to improve the quality of the guide task, allowing the users move more freely without any restrains. The sociable guide robot should not ask the users to follow it and maintain a proper distance with it. Instead, the robot should understand the will of the users and adapt to their motions. It is realized by environment understanding and controlling the robot by integrated potential field method, in which special potentials are generated from sub-goal and users. From the experimental results, the robot can adapt to the users' motions. Moreover, future works of this research is discussed before applyingto real environments. Verification experiments in public places are needed. The guide robot can be more considerate by analyzing the mental states of the users to predict intentions of the users.; 2016
• 国立国会図書館永続的識別子: info:ndljp/pid/10370800; https://dl.ndl.go.jp/pid/10370800
• コレクション（共通）: 障害者向け資料
• コレクション（障害者向け資料：レベル1）: テキストデータ
• コレクション（個別）: 国立国会図書館デジタルコレクション > デジタル化資料 > 博士論文; https://dl.ndl.go.jp/collections/A00014
• 収集根拠: 博士論文（自動収集）
• 受理日（W3CDTF）: 2017-07-03T04:10:06+09:00
• 作成日（W3CDTF）: 2017-04-24
• 記録形式（IMT）: application/pdf
• オンライン閲覧公開範囲: 国立国会図書館内限定公開
• デジタル化資料送信: 図書館・個人送信対象外
• 遠隔複写可否（NDL）: 可
• 掲載誌（URI）: http://id.nii.ac.jp/1438/00008504/
• 連携機関・データベース: 国立国会図書館 : 国立国会図書館デジタルコレクション; https://dl.ndl.go.jp

目次

2023-12-07 再収集