Constrained RGBD-SLAM - CEA - Commissariat à l’énergie atomique et aux énergies alternatives Accéder directement au contenu
Article Dans Une Revue Robotica Année : 2020

Constrained RGBD-SLAM

Résumé

SUMMARY This paper introduces a new RGBD-Simultaneous Localization And Mapping (RGBD-SLAM) based on a revisited keyframe SLAM. This solution improves the localization by combining visual and depth data in a local bundle adjustment. Then, it presents an extension of this RGBD-SLAM that takes advantage of a partial knowledge of the scene. This solution allows using a prior knowledge of the 3D model of the environment when this latter is available which drastically improves the localization accuracy. The proposed solutions called RGBD-SLAM and Constrained RGBD-SLAM are evaluated on several public benchmark datasets and on real scenes acquired by a Kinect sensor. The system works in real time on a standard central processing units and it can be useful for certain applications, such as localization of lightweight robots, UAVs, and VR helmet.
Fichier non déposé

Dates et versions

cea-04320509 , version 1 (04-12-2023)

Identifiants

Citer

Sylvie Naudet-Collette, Kathia Melbouci, Vincent Gay-Bellile, Omar Ait-Aider, Michel Dhome. Constrained RGBD-SLAM. Robotica, 2020, 39 (2), pp.277-290. ⟨10.1017/S0263574720000363⟩. ⟨cea-04320509⟩
13 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More