Real-Time Synchronized Positioning and Dense Map Construction from Feature-Based ORB-SLAM3

With the development of SLAM technology, many excellent SLAM algorithms have emerged, categorized into visual and laser SLAM, among others, based on the application scenario. This paper adopts the ORB-SLAM3 algorithm framework from visual SLAM. It optimizes the sparse point cloud data output by this algorithm, integrating reconstruction algorithms to ultimately produce visualized 3D dense point cloud data. The implementation primarily utilizes ORB-SLAM3's RGB-D mode. The semantic maps generated from further processing of this data can be applied in the field of augmented reality.