Learning-based Pose Coordination of Multirotor UAV Networks: Theory Development and Experimental Validation

2020 Fiscal Year Final Research Report

• Project/Area Number: 18K13775
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 21040:Control and system engineering-related
• Research Institution: Meiji University (2020); Tokyo Institute of Technology (2018-2019)
• Principal Investigator: Ibuki Tatsuya 明治大学, 理工学部, 専任講師 (30725023)
• Project Period (FY): 2018-04-01 – 2021-03-31
• Keywords: 制御工学 / 協調制御 / 機械学習 / ドローンネットワーク

Outline of Final Research Achievements

The objective of this project is to develop new pose coordination control theory incorporating a machine learning mechanism for multirotor UAV networks. Two topics were mainly tackled: (1) distributed pose coordination; and (2) learning-based pose control of multirotor UAVs. Especially, distributed optimization-based collision avoidance techniques were newly proposed in topic (1), which were published in international journals and conference proceedings. In topic (2), new pose control laws based on Gaussian processes were presented, where Gaussian processes estimated unknown dynamics of a multirotor UAV or unknown working environments. Experimental validation was also presented for both topics.

Free Research Field

制御・システム工学

Academic Significance and Societal Importance of the Research Achievements

モノやヒトのネットワーク化が目覚ましく進歩している現代においても，個々のモバイルロボットの自律分散的な制御によるロボット群の効率的な制御の実現には未だ課題が多く，一般的な技術として普及されていない．本研究はこの技術の根幹となるモバイルロボット群の分散型位置・姿勢協調制御問題に取り組み，さらに新たに機械学習機構を組み込んだ新規の制御手法を提案している．研究成果の学術的意義は多数の論文採録という形で既に認められている一方で，提案した分散型協調制御手法や機械学習機構の組み込み手法が，モバイルロボット群がますます活躍する社会の実現を目指した研究の今後の発展に寄与すると期待する．