Departmental Bulletin Paper 二自由度ロボットアームによる アフォーダンスの発見と獲得 :物理的性質の違いに基づいた分別タスクへの適用

長神, 雄三

Our research aimed to develop a new method for robot perception based on that of human beings. In this research, as a typical example, we employ a task involving distinguishing objects, and wedemonstrated the importance of motion in utilizing real-world properties. We focused on a learning mechanism to find usable real-world properties and to identify the motion that would enable us to utilize these properties. We employed shaking motion as an example, and we proposed a new learning procedure. Experiments were conducted, and we demonstrated that the difference of object could be distinguished autonomously by the proposed simple procedure.Key Words : robot, affordance, learning, artificial intelligence, perception

Number of accesses :  

Other information