Solving the problem of autonomous navigation using underlying surface models tolerant to survey conditions
А И Гаврилов, K. V. Parfentiev · AIP conference proceedings · 2021
The problem of autonomous navigation becomes more relevant in the aerospace industry. To solve it successfully, it is necessary to build underlying surface models tolerant to survey conditions. Such models can serve as a basis for solving wide variety of tasks, such as building topographic map of the terrain, navigating the underlying surface, automatic landing of the aircraft. The main problem with building underlying surface models according to monitoring data in the visible spectrum are changes in the survey conditions, such as time of year, time of day, various weather events. The problem can be solved by using algorithms of integration of data collected in various frequency bands and image preprocessing. The paper looks at the task of using data obtained through remote sensing of the underlying surface in optical and radar ranges with the purpose of determining navigational parameters and binding objects to space axes. The paper also considers different approaches to image preprocessing. Comparative studies of existing methods in this sphere were carried out. The algorithm for conversion of digital images into a tolerant to survey conditions form was suggested and realized. We developed the structure and software implementation of the system for determining geographical coordinates using preprocessed underlying surface image. Effectiveness of the suggested algorithms and software was validated by solving the problem of determining positions on images obtained with the help of satellite, radar, and aerial photography under different environmental and lighting conditions.