Socially Aware Robot Navigation Framework: Automatic Detecting and Autonomously Approaching People in Unknown Dynamic Social Environments

In this study, we propose a socially aware navigation framework that enables a mobile robot to automatically detect humans and navigate autonomously and safely to approach the detected humans. To do that the proposed framework make uses of a real-time appearance-based mapping (RTAB-Map) technique, which is RGB-D SLAM approach. The RTAB-Map method is based on the global loop closure detector with real-time constraints to generate 3D point clouds of the surrounding environment and create a 2D occupancy grid map for navigation process. The proposed framework also applies a frontier-based exploration approach, consisting of a flexible framework to decide a proper goal for mapping and exploring. Finally, the YOLO model is adopted to detect the human during the exploration process of the mobile robot. The simulation result illustrates that the mobile robot equipped with the proposed framework is able to automatically detect the human and navigate autonomously and safely to the detected humans in unknown dynamic environments