Simultaneous Localization and Mapping (SLAM) algorithms require huge computational power. Most of the state-of-the-art implementations employ dedicated computational machines which in most cases are off-board the robotic platform. In addition, as soon as the environment become large, the update rate of such algorithms is no more suitable for real-time control. The latest implementations rely on visual SLAM, adopting a reduced number of features. However, these methods are not employable in environments with low visibility or that are completely dark. We present here a SLAM algorithm designed for mobile robots requiring reliable solutions even in harsh working conditions where the presence of dust and darkness could compromise the visibility conditions. The algorithm has been optimized for embedded CPUs commonly employed in light-weight robotic platforms. In this paper the proposed algorithm is introduced and its feasibility as SLAM solution for embedded systems is proved both by a simulated and a real testing scenario.
A Lightweight SLAM algorithm for indoor autonomous navigation
TRIPICCHIO, Paolo;UNETTI, Matteo;GIORDANI, Nicola;AVIZZANO, Carlo Alberto;SATLER, MASSIMO
2014-01-01
Abstract
Simultaneous Localization and Mapping (SLAM) algorithms require huge computational power. Most of the state-of-the-art implementations employ dedicated computational machines which in most cases are off-board the robotic platform. In addition, as soon as the environment become large, the update rate of such algorithms is no more suitable for real-time control. The latest implementations rely on visual SLAM, adopting a reduced number of features. However, these methods are not employable in environments with low visibility or that are completely dark. We present here a SLAM algorithm designed for mobile robots requiring reliable solutions even in harsh working conditions where the presence of dust and darkness could compromise the visibility conditions. The algorithm has been optimized for embedded CPUs commonly employed in light-weight robotic platforms. In this paper the proposed algorithm is introduced and its feasibility as SLAM solution for embedded systems is proved both by a simulated and a real testing scenario.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
