Semantic Mapping based on ORB-SLAM and YOLO in indoor scenes

Abstract

Simultaneous localization and mapping (SLAM) technology has always been an important area of robotics research. With the improvement of hardware accuracy and reliability of depth cameras, the application of depth cameras as the main sensor for indoor navigation robots has become a recent trend. On the other hand, camera sensors have become the primary means of obtaining environmental information due to their advantages such as low cost and high scene recognition capability. The objective of this research is to generate a map that can be applied to voice navigation by integrating the semantic information of objects detected by the object detection algorithm YOLO into the navigation map created by Visual SLAM. This report describes a mapping system based on object detection and Visual SLAM.