The article discusses a method for accurate loop closure detection in Semantic Simultaneous Localization and Mapping (SLAM) using 3D semantic covisibility graphs. Loop closure helps correct errors that accumulate in SLAM during a robot’s motion. The proposed approach, SmSLAM+LCD, integrates high-level 3D semantic information with low-level feature information for accurate loop closure detection and drift reduction. This method addresses the challenge of distinguishing similar scenarios or environments in previous geometric feature-based methods. The effectiveness of this approach is demonstrated through testing results.
Publication date: 22 Nov 2023
Project Page: https://ieeexplore.ieee.org/document/9677345
Paper: https://arxiv.org/pdf/2311.12245
