Solution de génération de formes 2D d’objets basée sur des connaissances préalables pour la cartographie sémantique
Résumé
This study presents an RGBD-based real-time semantic mapping solution. We focus on presenting a novel association approach to generate 2D semantic object shapes based on prior knowledge. We evaluated our approach in an office environment using the mobile robot MIR. Our experimental results and comparison with an existing approach show that our proposal generates maps that are very close to the ground truth.