Simulation of UAV-based indoor mapping for multi-story building using a RGB-D SLAM solution

During post-disaster rescue operations in buildings affected by events such as earthquakes, searching in unknown indoor environments where GNSS signals are blocked becomes inevitable. For missions of this nature that potentially endanger the lives of search and rescue personnel, autonomous reconnaissance of unfamiliar surroundings can be achieved through the deployment of unmanned aerial vehicles (UAV) utilizing Simultaneous Localization and Mapping (SLAM) techniques. However, for multi-story buildings, the existing methods suffer from limited perceptual range between floors due to vertical height differences, thus limiting their applicability. To address this challenge and meet the need for continuous traversal and exploration between building levels, a perception approach integrating ultrasonic and RGB-D is proposed for inter-floor localization. This study constructs multi-story buildings using the Gazebo simulator to evaluate the proposed methodology, enabling the generation of dense point cloud maps and occupancy grid maps of indoor environments.