Navigation data extraction from monocular camera images during final approach

Abstract

Fault tolerant flight control systems require multiple independent sources of navigation data, especially in the case of low altitude maneuvers. During final approach, the relative position and orientation of the aircraft can be computed based on monocular camera images in which the runway presents. However, this navigation sensor has different error characteristics compared to ILS or GNSS. This paper presents the first steps towards a runway detection method with subpixel accuracy and collects the main possible bottlenecks of vision-aided navigation.