UAV and Soft Computing Methodology for Monitoring Landslide Areas (Susceptibility to Landslides and Early Warning)

Abstract

In this work, we created a map of the susceptibility to landslides in GIS environment using neural network, Analytical Hierarchy Process (AHP) multicriteria analysis method and fuzzy methodology, producing five categories (levels) of risk. Subsequently, starting from this map, we identified (fuzzy methodology) the areas of the road’s network most exposed to landslide risk also using remote sensing techniques (classification and segmentation techniques) overlapped on the street map. This system therefore provides us the level of attention that affects the transport infrastructure investigated (a higher level of attention corresponds to a higher level of landslide risk). Once the risk map for a large area was identified, we focused on local monitoring of a part of it automatically selected by the GIS. The monitoring of this area was carried out through an innovative system (made by us) that allows to monitor landslide risk areas and to study landslide phenomena through the use of Unmanned Aerial Vehicles (UAVs). Specifically, with this innovative solution, data are acquired thanks to an automated system of UAVs and wireless charging platforms (capable to acquired, to transmit and to store data); subsequently, the acquired data are stored automatically in a special platform that allows us to create the point cloud and 3D models of the investigated area (which in turn they are superimposed on the digital models created in previous monitoring), also allowing the creation of the land mass displacement’s sequence in a video. Finally, in relation to early warning, the system allows civil protection to be warned in the event of a landslide risk (start of new landslides or continuation of landslides that have already begun) which in this way will be able to warn the population also through social media.