Airport markings recognition for automatic taxiing

Taxiing within the airport of Unmanned Aerial Vehicles (UAV) is currently based on front camera video provided to the remote pilot and GPS information. The latter introduces a systematic and, most tricky, random error that cannot allow the complete automatic taxiing. This paper presents a novel approach where image processing allows a GPS signal rectification, making use of automatic recognition of airport signs and markings. Runaway paths as well as taxiway location signs are identified and correlated to the GPS information. Several approaches had made use of Hough transform to find the runway for landing or take-off without cooperation between position and optical systems. The aim of this work is to create autonomous taxiing based on GPS information corrected by automatic recognition of airport signs, especially useful (but not limited) to UAV. Processing requirements have been identified and tests have been performed on videos recorded during taxiing at civil airports making use of UAVs remotely driven.