歩行者混雑状況下におけるメタ強化学習を用いた移動ロボットナビゲーション

抄録

Mobile robots used in our daily lives must be able to move safely and adaptively in a variety of situations, such as varying numbers of pedestrians, different walking speeds, and varying levels of crowd density. In this paper, we propose crowd-aware robot navigation methods using meta reinforcement learning. We verify whether our methods can adapt to multiple scenario through meta test.