Design and experimental validation of advanced guidance and navigation algorithms for Unmanned Aerial Vehicles

Autonomous navigation in GNSS-denied environments remains a critical challenge for Unmanned Aerial Vehicles (UAVs), particularly for indoor applications such as infrastructure inspection and disaster first response. This study presents the design, simulation, and experimental validation of a navigation system based on a low-cost 2D Light Detection and Ranging (LiDAR) sensor and Simultaneous Localization and Mapping (SLAM) algorithm. Utilizing the Robot Operating System (ROS)2 framework and the PX4 flight control architecture, the slam_toolbox package was implemented to provide real-time pose estimation and mapping. Extensive simulations in the Gazebo environment were conducted to evaluate the system’s performance under varying LiDAR sample rates (200, 400, and 800 samples/revolution) and map resolutions both for hovering and navigation. The results demonstrate that accurate hovering can be achieved with a mean absolute error within tens of centimeters, comparable to RTK GNSS precision in simulation. However, it was identified a critical trade-off between map fidelity and computational cost, where high-resolution configurations (0.01 m) triggered hardware-limited latency and system instability. The study emphasizes the necessity of precise external vision delay tuning to ensure stability. It is then proposed a deployment strategy on a Raspberry Pi 5 companion computer to bridge the gap between simulation and real-world indoor flight. Finally, the Experimental Validation was performed in the flight arena of the CICLoPE facility and the results are presented and discussed, demonstrating that the proposed architecture enables stable and accurate state estimation, confirming the effectiveness of the implemented framework for UAV operations in indoor environments.