Multi-information Based Safe Area Step Selection Algorithm for UAV’s Emergency Forced Landing

Abstract

In order to solve the problem of forced landing in emergency for Unmanned Aerial Vehicles (UAVs), a multi-information based algorithm for selecting forced landing area by step is proposed. In order to extract the slowly varying edges and weak edges in an aerial image, this algorithm adopted improved edge detection method to detect landing area without obstacles. To select the detected safe areas which are suitable for landing in terms of size and shape, four masks with adequate coverage were designed. The elevation data of the areas were acquired to analyze its terrain. By extracting features of color and texture based on Gray Level Co-Occurrence Matrix (GLCM), fast classification and recognition of landing areas was carried out based on Support Vector Machine (SVM) classifier. Simulation results show that the algorithm, comparing with Bayesian classifier, presents a more fast and accurate classification and selection of the landing areas, fulfilling the demand of forced landing for UAVs in emergency.