Fast Adaptation of Reinforcement-Learned Robot Navigation Skills to Human Preferences - Naver Labs Europe