Introducing LiDAR Point Cloud-based Object Classification for Safer Apron Operations

Current procedures for conventional and remote airport ground control still rely on the direct (camera-) view. Despite further support by different Radar applications occasional shortcomings in the awareness of the responsible controllers may occur, particularly under adverse weather conditions, giving rise to capacity backlogs, incidents and accidents. As Laser scanners and computer vision algorithms have reached new performance levels in recent years, we proposed a novel concept for complete and independent airport apron surveillance based on LiDAR 3D point data. In this paper we extend our object detection/segmentation technique by addressing object classification in LiDAR 3D scans. We hereby enable LiDAR`s unique capability to classify noncooperative objects by means of a single sensor and learned model knowledge. Our technique was able to classify and to estimate the poses of an Airbus A319-100 and a Boeing B737-700 parked on the airport apron. In the future we will enhance our classification technique to a wider range of objects including moving ground vehicles and pedestrians. KeywordsLiDAR, Laser scanning, 3D point cloud, airport ground surveillance, apron control, aircraft classification, pose estimation