Metric, dense and real-time 3D Reconstruction for drone autonomous navigation
Reconstruction 3D dense, métrique et temps réel pour la navigation autonome de drone
Résumé
Simultaneous Localization And Mapping (SLAM) research have reach maturity allowing state of the art algorithm to build a 3D dense and metric reconstruction in real-time. However, it becomes more challenging in drone navigation context where we favor passive sensors in monocular configuration and need to tackle fast motions or illumination changes. Research on Deep learning monocular depth estimation and event cameras present interesting advances to resolve these challenges.