Localization of Aerial Vehicles performing Aggressive Maneuvers in GPS-unreliable Environments Using Vision information and Adaptive State Estimators
Antonio C. B. Chiella1; Bruno O. S. Teixeira1; Guilherme A. S. Pereira1
1 Universidade Federal de Minas Gerais
Resumo
This paper considers the problems of localization and flight path reconstruction of aerial vehicles in GPS-unreliable environments. Our motivation comes from aerobatics competitions and aerial races, where the GPS signal can be lost during the execution of aggressive maneuvers, or damaged due to the strong interference caused by the digital television transmitters installed on the aircraft. Despite these problems, a good path estimate in such situations would be helpful for the pilot to improve the flight performance. In this paper, we investigate two adaptive filters and smothers based on the unscented Kalman filter/smoother. Aiming at a better path estimate, we also propose the inclusion of vision information obtained from the onboard camera. Experimental data collected from an airplane flying in a session of the Red Bull Air Race is used to illustrate the performance of the methods. Substantial improvements are observed when the adaptive algorithms and camera information are used for the vehicle localization and the reconstruction of its path.
Palavras-chave: Adaptive Kalman filtering; Localization; Flight path reconstruction