深層強化学習を用いた群ロボットの協調行動の獲得に関する検討

カテゴリ: 部門大会

論文No: TC13-5

グループ名: 【C】2022年電気学会電子・情報・システム部門大会

発行日: 2022/08/24

タイトル(英語): A Study on Behavior Acquisition in Multi-robots System by Deep Reinforcement Learning

著者名: 福島 英（出雲村田製作所）,曽田 涼介（松江工業高等専門学校）,堀内 匡（松江工業高等専門学校）

著者名(英語): Akira Fukushima (Izumo Murata Manufacturing Co., Ltd.),Ryosuke Sota (National Institute of Technology, Matsue College),Tadashi Horiuchi (National Institute of Technology, Matsue College)

キーワード: 深層強化学習|群ロボット|行動獲得|視覚ベース|Deep Reinforcement Learning|Multi-robots System|Behavior Acquisition|Vision-based

要約(日本語): In this research, we apply deep reinforcement learning methods such as DQN and Soft-Actor Critic to multi-robot environment, in order to realize that multiple mobile robots acquire the behavior to overtake other robots or to follow other robots. These tasks are more difficult in multi-robot environment than in a single robot environment, due to the influence by the actions of other robots. The robots have cameras and distance sensors attached to themselves, and all three robots move in the same direction. In this study, there are two learning tasks to be realized. First task is to learn the behavior of a robot overtaking  other robots. Second task is to learn the behavior of a robot following the other robots.