This paper presents an original approach for autonomous navigation based on RGB-D data and known 3D markers, where the basic concept is to detect and recognize the markers and then to use them for a straightforward pose estimation solution. The developed algorithms can allow a quadrotor to autonomously fly in (cooperative) GPS denied environments and/or when there is no natural or artificial illumination of the scene, by following a predetermined path consisting of successive targets having a well defined shape and/or color. Algorithms for target detection and recognition based on depth data are described which are optimized for real time use, paying particular attention to the on-board computational load. Experimental tests have been carried out by integrating a RGB-Depth sensor (ASUS Xtion Pro Live) on-board a custom-built quadrotor. First results confirm the potential of the proposed approach. The technique can be applied to different types of unmanned aerial vehicles (UAVs), as well as unmanned ground vehicles (UGVs).
RGB-D camera-based quadrotor navigation in GPS-denied and low light environments using known 3D markers / Vetrella, AMEDEO RODI; Savvaris, Al; Fasano, Giancarmine; Accardo, Domenico. - (2015), pp. 185-192. (Intervento presentato al convegno 2015 International Conference on Unmanned Aircraft Systems, ICUAS 2015 tenutosi a Denver Marriott Tech Center, usa nel 2015) [10.1109/ICUAS.2015.7152290].
RGB-D camera-based quadrotor navigation in GPS-denied and low light environments using known 3D markers
VETRELLA, AMEDEO RODI;FASANO, GIANCARMINE;ACCARDO, DOMENICO
2015
Abstract
This paper presents an original approach for autonomous navigation based on RGB-D data and known 3D markers, where the basic concept is to detect and recognize the markers and then to use them for a straightforward pose estimation solution. The developed algorithms can allow a quadrotor to autonomously fly in (cooperative) GPS denied environments and/or when there is no natural or artificial illumination of the scene, by following a predetermined path consisting of successive targets having a well defined shape and/or color. Algorithms for target detection and recognition based on depth data are described which are optimized for real time use, paying particular attention to the on-board computational load. Experimental tests have been carried out by integrating a RGB-Depth sensor (ASUS Xtion Pro Live) on-board a custom-built quadrotor. First results confirm the potential of the proposed approach. The technique can be applied to different types of unmanned aerial vehicles (UAVs), as well as unmanned ground vehicles (UGVs).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
