CCTV-Informed Human-Aware Robot Navigation in Crowded Indoor Environments

Abstract

Mobile robot navigation in crowded indoor environments is a challenging task due to the limited sensing capabilities of onboard sensors. In this study, we propose a mobile robot navigation framework that utilizes external CCTV data to address the limitations of local sensors in a crowded environment. This approach enables mobile robots to navigate safely and efficiently in complex environments by encapsulating human movements from CCTVs to anticipate the human impact on the unclear navigational trajectory of our robot and devise human-aware paths that mitigate collision risks and minimize social intrusions. Further, we integrate a deep reinforcement learning (DRL) algorithm into a generated global path to fine-tune robotic navigation in human-populated areas, enabling the robot to learn efficiently and socially acceptable navigation compared to methods based solely on local sensors. Our experiments further validate the efficiency of using CCTVs to supplement robots with constrained sensing across varied sensor capabilities and CCTVs configurations.