LiDAR localization improves accuracy with new scene feature method
ReLoc: Rethinking Scene Coordinate Regression Architecture for Robust Outdoor LiDAR-based Localization
Computer Vision and Pattern Recognition
Summary
Locating a device outdoors using LiDAR sensors can be tricky without a detailed 3D map. The authors found that current methods mix up features from areas that look similar and include noisy data from moving things, which makes locating less reliable. They created ReLoc, a new approach that better understands the overall scene and focuses on stable, meaningful details. Tests show this makes pinpointing locations more accurate and still fast enough to use in real time.
What this means in practice
- •For autonomous vehicle engineers: Use ReLoc to improve vehicle position tracking accuracy outdoors without detailed 3D maps.$Commercial implications: Enables more reliable and marketable navigation systems for autonomous vehicles by improving LiDAR localization robustness.
- •For robotics system developers: Integrate an attention-based feature method to enhance localization stability in dynamic outdoor environments.
Authors
Heejoon Moon, Yurim Cho, Je Hyeong Hong
Abstract
Scene Coordinate Regression (SCR) has recently emerged as a promising approach for LiDAR-based localization, achieving accurate localization without requiring an explicit 3D map. Despite their effectiveness, existing SCR methods rely on scene classification-based global embedding that struggles to provide fine-grained discrimination among nearby locations. Moreover, their reliance on uniform sampling of local features during training assigns equal importance to all points, thereby inadvertently propagating features from dynamic objects or unstable regions and potentially degrading training stability. In this paper, we present ReLoc, a revamped SCR architecture that can effectively address these limitations. First, we redesign the global embedding module by combining learnable context tokens with a feature aggregator to capture richer and more discriminative scene context. Second, we introduce an attention-based local feature enhancement module to mitigate the impact of noisy local features while encouraging context-consistent structures, yielding more robust local feature representations. Experimental results on two large-scale outdoor datasets demonstrate that our approach achieves state-of-the-art accuracy over previous SCR-based methods while maintaining real-time inference performance.