Automatic Map Update for Autonomous Navigation in GNSS-Denied Environments

This thesis addresses the problem of automatic map updating in dynamic environments, with a particular focus on indoor airport environments, where even small variations in the structural layout can compromise the autonomous navigation capabilities of mobile robots. After an initial analysis of the ROS 2 framework, the slam_toolbox package, and the main Simultaneous Localization and Mapping (SLAM) algorithms, a methodology is proposed for identifying and classifying differences among maps acquired at different time instants. The experimental validation was carried out in simulated environments using Gazebo and RViz, by generating custom maps and developing Python algorithms based on the OpenCV and NumPy libraries for identifying discrepancies. Furthermore, a system integrated into the ROS 2 navigation stack in a decentralized environment was implemented, responsible for map management, pose estimation, and data storage through specific nodes. The results confirm the feasibility of automatic map updating, paving the way for the development of more robust and adaptable autonomous robotic systems suitable for real-world scenarios.