Autonomous Vision-Based Safe Area Selection Algorithm for UAV Emergency Forced Landing

Abstract

In order to solve forced landing for Unmanned Aerial Vehicle (UAV) in emergency situations, an autonomous vision-based algorithm to select safe forced landing area was proposed. To detect the slowly changing edges and weak edges, the algorithm improved the canny operator to detect safe landing area without obstacles. Then eroding and dilating close operation was adopted to select the area. To analyze its terrain, the DEM data was acquired. By extracting features of interesting area, fast classification and recognition of landing area based on Bayesian theory was carried out. The algorithm presents a fast and accurate result on the forced landing area selection for UAVs.