RALoc: Enhancing Outdoor LiDAR Localization via Rotation Awareness

Yuyang Yang1,2*, Wen Li1.2*, Sheng Ao1.2, Qingshan Xu3, Shangshu Yu4, Yu Guo1,2, Yin Zhou5, Siqi Shen1,2, Cheng Wang1,2†
1Fujian Key Laboratory of Sensing and Computing for Smart Cities, Xiamen University 2Key Laboratory of Multimedia Trusted Perception and Efficient Computing Ministry of Education of China, Xiamen University 3Nanyang Technological University 4Northeastern University 5GAC R&D Center
Paper Supp Data Code

Abstract

LiDAR localization is a fundamental task in autonomous driving and robotics. Scene Coordinate Regression (SCR) exhibits leading pose accuracy, achieving impressive results in learning-based localization. We observe that the real- world LiDAR scans captured from different viewpoints usu ally result in the catastrophic collapse of SCR. However, ex isting LiDAR localization methods have largely overlooked the issue of rotation sensitivity in SCR. In this paper, we present RALoc, an outdoor LiDAR localization method with rotation awareness to achieve accurate localization. The key to our approach is to design a Point Cloud Canon icalization module, which leverages a powerful equivari ant key feature aggregation to transform the input LiDAR scan towards a consistent orientation, effectively eliminat ing the adverse effects of rotation. This proposed module has promising scalability and can be seamlessly integrated with the existing LiDAR localization network. Moreover, we propose the Bidirectional LiDAR Localization (BiLiLo) dataset as a benchmark to evaluate the performance of var ious methods in large outdoor scenes with significant rota tion changes. Extensive experiments show that RALoc sig nificantly improves localization performance in scenarios with large rotation changes, and also achieves competitive performance in the Oxford Radar RobotCar dataset.

teaser_image

Dataset

Log in using the username "Guest" and the password "guest_CMD" to download the dataset.

For any questions, please email yangyuyang_01@163.com.

BibTeX

@inproceedings{raloc,
          title={RALoc: Enhancing Outdoor LiDAR Localization via Rotation Awareness},
          author={Yang, Yuyang and Li, Wen and Ao, Sheng and Xu, Qingshan and Yu, Shangshu and Guo, Yu and Zhou, Yin and Shen, Siqi and Wang, Cheng},
          booktitle=ICCV,
          year={2025}
        }