Robust IR-Based Pose Estimation for Precision VTOL Aircraft Landing in Urban Environments

This paper presents a novel pose estimation frame-work that provides high-accuracy localization for vertical take-off and landing aircraft in settings where GPS is unreliable or unavailable. The proposed framework utilizes a sparse constellation of infrared lights that can be robustly identified and associated in the presence of outliers and occlusions, making it suitable for use in realistic urban environments. This is enabled by constellation designs that exploit properties of invariance under projective transformations. Flight test results demonstrate that the framework is capable of running in real-time at speeds of over 30 Hz while providing pose information at decimeter-level accuracy at ranges of over 100 m from the landing site.