Skip to Main Content
In this paper a vision-based system for safe autonomous landing of a helicopter-based Unmanned Aerial Vehicle (UAV) is presented. The remote user selects target areas from high resolution aerial or satellite images. These areas are tracked by a feature-based image matching algorithm that identifies natural landmarks and gives feedbacks for control purposes. The main novelty of the proposed approach is on the use of textures for terrain classification before landing, in addition to the optical flow procedures used in the system described in. The new procedure allows the UAV to identify suitable landing areas through a comparison between the image sequences taken by the onboard camera and a database of known textures, somehow representing the aspect of safe grounds (e.g., grass or gravel). The adoption of the two procedures aims to make autonomous landing safer by considering terrain analysis from two different perspectives.