Analysis of the applicability of visual navigation methods in arctic conditions

Abstract

The article discusses the features of the use of autonomous navigation algorithms for mobile robotics in the Arctic. The main difficulty of the work of visual navigation methods in such conditions is that large areas of frames are occupied by a uniform snow cover, which greatly complicates the search for characteristic points and their comparison, as a result, it becomes impossible to achieve the necessary localization accuracy and map construction. The paper describes ways to improve the quality of visual navigation in arctic conditions - determining the skyline, which allows a more detailed view of the area of movement of the robot, not taking into account variable cloud cover and finding characteristic points not only on large objects on the horizon, but also on snow; adaptive contrast equalization, which is used to display such imperceptible features of a snowy surface as ribbing left by the wind, which makes it possible to distinguish features even on a visually uniform plane; the use of scale-invariant features; a special approach to the selection of key frames.