РАЗРАБОТКА АЛГОРИТМА НАВИГАЦИИ И ФОРМИРОВАНИЕ БЕЗОПАСНЫХ ТРАЕКТОРИЙ АВТОНОМНОГО ПОЛЕТА БЕСПИЛОТНОГО ЛЕТАТЕЛЬНОГО АППАРАТА В УСЛОВИЯХ МНОГОЧИСЛЕННЫХ ПОМЕХ И ПРЕПЯТСТВИЙ

Abstract

Relevance The problem of navigation of unmanned aerial vehicles (UAV) in monitoring forest ecosystems is considered. The relevance of this study lies in the development of navigation algorithms and the formation of safe trajectories for autonomous flight of UAVs in the face of numerous interference and obstacles. Aim of research Analysis of modern means of forest ecosystem monitoring. The objectives of the research project are: 1. When solving the navigation problem, it is proposed to use «machine vision» technologies. These technologies make it possible to find and localize dangerous scene objects on the current image in order to generate signals for controlling the movement of the UAV along safe trajectories. 2. Development of an algorithm for segmenting the image of the environment based on a statistical analysis of the histograms of the brightness of image areas. This algorithm makes it possible to perform airspace zoning according to the hazard level using the introduced set of textural features. 3. Evaluation of the efficiency of the forest image segmentation algorithm based on the calculation of statistical texture characteristics. An algorithm for forming the trajectory of low-flying UAVs to avoid danger is presented. Research method Determination of texture features based on statistical analysis of histograms of brightness of image areas. Results The algorithm for forming the trajectory of low-flying UAVs in conditions of numerous interference and obstacles makes it possible to evaluate potentially dangerous sections of the flight trajectory in order to develop recommendations for adjusting coordinates and movement parameters that ensure the required level of safety, including automatic maneuvering to avoid danger. During this research project, a navigation algorithm was developed, as a result of which the modeling results presented in the work confirm the high efficiency of the proposed algorithm for segmenting images of forest areas obtained from cameras or laser-location methods in order to form the trajectory of low-flying UAVs in tasks of monitoring forest ecosystems.