EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

R2SCAT-LPR: Rotation-Robust Network with Self- and Cross-Attention Transformers for LiDAR-Based Place Recognition

Weizhong Jiang, Hanzhang Xue, Shubin Si, Liang Xiao, Dawei Zhao, Qi Zhu, Yiming Nie, Bin Dai

Remote Sensing · 17.03.2025 · 17 (6) · 1057

DOI: 10.3390/rs17061057 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 17.03.2025. Verfasst von Weizhong Jiang und 7 weiteren Autor:innen. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Defense Innovation Institution.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

LiDAR-based place recognition (LPR) is crucial for the navigation and localization of autonomous vehicles and mobile robots in large-scale outdoor environments and plays a critical role in loop closure detection for simultaneous localization and mapping (SLAM). Existing LPR methods, which utilize 2D bird’s-eye view (BEV) projections of 3D point clouds, achieve competitive performance in efficiency and recognition accuracy. However, these methods often struggle with capturing global contextual information and maintaining robustness to viewpoint variations. To address these challenges, we propose R2SCAT-LPR, a novel, transformer-based model that leverages self-attention and cross-attention mechanisms to extract rotation-robust place feature descriptors from BEV images. R2SCAT-LPR consists of three core modules: (1) R2MPFE, which employs weight-shared cascaded multi-head self-attention (MHSA) to extract multi-level spatial contextual patch features from both the original BEV image and its randomly rotated counterpart; (2) DSCA, which integrates dual-branch self-attention and multi-head cross-attention (MHCA) to capture intrinsic correspondences between multi-level patch features before and after rotation, enhancing the extraction of rotation-robust local features; and (3) a combined NetVLAD module, which aggregates patch features from both the original feature space and the rotated interaction space into a compact and viewpoint-robust global descriptor. Extensive experiments conducted on the KITTI and NCLT datasets validate the effectiveness of the proposed model, demonstrating its robustness to rotation variations and its generalization ability across diverse scenes and LiDAR sensors types. Furthermore, we evaluate the generalization performance and computational efficiency of R2SCAT-LPR on our self-constructed OffRoad-LPR dataset for off-road autonomous driving, verifying its deployability on resource-constrained platforms.

Förderung

  • Defense Innovation Institution

Zitieren

APA

Jiang, W., Xue, H., Si, S., Xiao, L., Zhao, D., Zhu, Q., Nie, Y., & Dai, B. (2025). R2SCAT-LPR: Rotation-Robust Network with Self- and Cross-Attention Transformers for LiDAR-Based Place Recognition. Remote Sensing, 17(6), 1057. https://doi.org/10.3390/rs17061057

Vancouver

Jiang W, Xue H, Si S, Xiao L, Zhao D, Zhu Q, et al. R2SCAT-LPR: Rotation-Robust Network with Self- and Cross-Attention Transformers for LiDAR-Based Place Recognition. Remote Sensing. 2025;17(6):1057. doi:10.3390/rs17061057

BibTeX
@article{Jiang2025,
  author = {Jiang, Weizhong and Xue, Hanzhang and Si, Shubin and Xiao, Liang and Zhao, Dawei and Zhu, Qi and Nie, Yiming and Dai, Bin},
  title = {R2SCAT-LPR: Rotation-Robust Network with Self- and Cross-Attention Transformers for LiDAR-Based Place Recognition},
  journal = {Remote Sensing},
  year = {2025},
  volume = {17},
  number = {6},
  pages = {1057},
  doi = {10.3390/rs17061057}
}
Kostenlos starten →