Many indoor mobile robots spreading in logistics and service applications rely on localization methods based on fixed reflectors and LIDARs. In order to remove the need to artificially place external landmarks, several research institutes and companies have been working on natural navigation for years. When performing natural navigation, a robot creates a map of the environment and then uses it for localization purposes. Since the map should accurately represent the environment, these methods are not much robust to the environmental changes that affect most real-world scenarios during long-term operations. Map updating methods have been developed to avoid recreating a new map from scratch when the old one becomes obsolete. Unfortunately, most of these methods are not robust to localization error or are expensive in terms of computational cost and required memory.
In this thesis, a novel LIDAR-based map updating algorithm for dynamic environments is proposed. The approach is validated through simulation, reporting promising results.