2A1-E18 タスク実行中のロボットの回避運動学習

概要

論文の詳細を見る
Due to the rapid advancement of the robot technology, application area of robot has been extended from industry to our daily-life environment. However, in such environment, the situation is irregularly changing and unpredictable, so that it is difficult for robots to execute the given tasks without accidental events. In this paper, as one of such problems, we take up the autonomous avoidance behavior from the unexpected obstacles while the robot is performing a given task. Especially we focus on the learning of a two-link manipulator to avoid an approaching object while it is standing up. Through two processes of the reinforcement learning, the robot can acquire the ability to perform the standing-up task and the avoidance action. The computer experiment showed that the proposed method gives satisfactory results.
一般社団法人日本機械学会の論文