UAV Autonomous Indoor Exploration and Mapping for SAR Missions: Reflections from the ICUAS 2022 Competition

摘要

The technological advancement in Unmanned Aerial Vehicles (UAVs) or drones and their deployment in real-life Search and Rescue (SAR) missions is imminent. We, therefore, present a perception-aware autonomous exploration framework aimed at performing vision-based target detection and collision avoidance with an Unmanned Aerial Vehicle (UAV). The UAV utilizes a depth camera for maneuvering and finding the target. The underlying indoor exploration approach considers autonomous collision-free navigation, as well as target detection with a ballistic ball payload delivery without a prior map. Moreover, the proposed method allows safe navigation in enclosed unknown areas congested with randomly positioned obstacles and target locations. Our underlined end-to-end system architecture integrates the proposed exploration strategy. Extensive simulation experiments, using several Key Performance Indicators (KPIs), showcase the effectiveness of the proposed Robot Operating System (ROS) framework in a simulated Gazebo environment under various parameter settings.