Reliable LiDAR Loop Detection through Structural Descriptors and Semantic Graph Matching
Yujie Tang, Sibo Zuo, Meiling Wang, Jianyu Dou, Jiahui Wang, Yufeng Yue
AI summary
Problem
Existing LiDAR loop closure methods struggle with reliability in varied environments: structural descriptors lack semantic awareness and suffer from sparsity and noise, while semantic graph methods fail in object-sparse scenes and struggle with ambiguous node matching.
Approach
The framework extracts semantic node graphs and global structural descriptors from point clouds, uses multi-scale features and maximum clique algorithms for robust node matching, and fuses both similarity scores via a cross-validation mechanism that adaptively weights them based on reliability cues.
Key results
- State-of-the-art F1 max and EP scores on KITTI and KITTI-360 datasets
- Adaptive cross-validation mechanism weighting semantic and structural scores via node richness and yaw discrepancy
- Multi-scale features and maximum clique outlier rejection for globally consistent node correspondences
- Verified improvement in SLAM trajectory accuracy and semantic map optimization on real-world campus data
Why it matters
Provides a robust, adaptable loop closure solution for LiDAR-based SLAM, enabling reliable long-term navigation and mapping in diverse outdoor environments.
Abstract
Outdoor loop closure detection is essential for mitigating accumulated drift in SLAM and generating a global consistent map. Semantic graph matching methods utilize object-level topology for distinctive scene representation but rely on environments with rich and distinguishable objects. Moreover, accurately matching nodes remains difficult due to ambiguities among same-class semantic nodes. These chal- lenges limit their effectiveness in varied road environments, highlighting the need for representations that are both robust and adaptable. To address this, we introduce SD-SGM, a novel loop closure detection framework combining the powerful context-adaptation capabilities of structural descriptors with the high-level semantic reasoning abilities of semantic graphs. Initially, we extract semantic graphs alongside global structural descriptors from point clouds. Distinctive local graph features are then used to generate candidate node pairs, and the maximal clique algorithm identifies correspondences that are globally consistent. The similarity scores of both methods are then evaluated and a cross-validation mechanism assesses their reliability and adaptively weights them. Extensive loop closure detection experiments on various datasets demonstrate that SD- SGM achieves state-of-the-art (SOTA) performance compared to strong baselines. Additionally, we verify its effectiveness in improving SLAM trajectory accuracy. We provide the code at: https://github.com/BIT-TYJ/SD-SGM.