C-LaV：面向天气鲁棒 LiDAR 地点识别的条件潜变量速度场去噪
C-LaV: Conditional Latent Velocity Field Denoising for Weather-Robust LiDAR Place Recognition
把条件流匹配当作潜空间去噪器，在冻结的 DINOv2 BEV 潜空间里用概率流 ODE 把雨雪雾潜变量确定性搬运到晴天的对应潜变量，再接 SALAD 聚合做检索；这套「冻结语义锚定潜空间 + 条件流匹配去噪」的范式可直接迁移到其他受天气/传感器退化影响的检索与定位任务。
原文依据 a conditional Flow Matching model learns a velocity field whose probability-flow ordinary differential equation (ODE) deterministically transports noisy latents toward their clear-weather counterparts
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基于 LiDAR 的地点识别对雨、雪和雾高度敏感，散射与衰减会扭曲几何结构和强度。本文以条件潜变量速度场（C-LaV）去噪来解决该问题，在检索之前恢复出对天气鲁棒的表示。单帧点云被投影为三通道鸟瞰图（BEV）图像，并由冻结的、基于 DINOv2 的 BEV Transformer 编码，从而获得一个跨天气条件共享、语义锚定的潜空间。在该流形上，条件流匹配模型学习一个速度场，其概率流常微分方程（ODE）确定性地将含噪潜变量输运至其晴天对应物。在去噪后的流形上，Sinkhorn 局部描述子聚合（SALAD）检测头生成紧凑的全局描述子，并以截断 Smooth-AP 损失进行优化。我们还建立了一个统一的恶劣天气基准，在 KITTI、NCLT 和 Boreas 数据集上采用 3 米帧间距与共享的评测阈值。在该协议下，C-LaV 在 NCLT 雪天场景上将 Recall@1 提升 17.5%，在 Boreas 上提升 21.5%，达到最优的天气鲁棒性。我们的数据集与代码将公开。
英文原文摘要 1237 字符
LiDAR-based place recognition is highly sensitive to rain, snow, and fog, where scattering and attenuation distort geometric structure and intensity. We tackle this problem with Conditional Latent Velocity Field (C-LaV) denoising, which restores weather-robust representations before retrieval. Single-sweep point clouds are projected into three-channel bird's-eye-view (BEV) images and encoded with a frozen DINOv2-based BEV transformer to obtain a semantically anchored latent space shared across weather conditions. On this manifold, a conditional Flow Matching model learns a velocity field whose probability-flow ordinary differential equation (ODE) deterministically transports noisy latents toward their clear-weather counterparts. From the denoised manifold, a Sinkhorn Aggregation of Local Descriptors (SALAD) head produces compact global descriptors optimized with a truncated Smooth-AP loss. We also establish a unified adverse-weather benchmark with 3 m frame spacing and shared evaluation thresholds across KITTI, NCLT, and Boreas datasets. Under this protocol, C-LaV improves Recall@1 by 17.5% on NCLT snow and 21.5% on Boreas, achieving state-of-the-art weather robustness. Our dataset and code will be publicly available.