Register Anything Model：可泛化且鲁棒的点云配准统一模型
原标题：Register Anything Model for Generalizable and Robust Point Cloud Registration
AI 摘要
该论文提出 Register Anything Model（RAM），一个统一的三维点云配准模型，可在室内、室外和物体等多领域数据上训练，并稳健估计 6-DoF 变换。针对跨域几何差异（场景尺度、传感器噪声、点密度、结构模式），作者引入几何感知归一化策略，将点云体素化并归一化到标准化几何空间，同时采用高效几何变换器建模局部几何一致性。实验表明，RAM 在 inlier ratio 和 registration recall 上达到 SOTA，且优于在各单独领域训练的模型，展现出更强的泛化能力。
正文节选
Register Anything Model for Generalizable and Robust Point Cloud Registration Abstract We present Register Anything Model (RAM), a unified model for generalizable 3D point cloud registration that robustly estimates 6-DoF transformations across diverse domains spanning sensors and environments. Existing registration methods are typically tailored to a single domain, with carefully tuned architectures and hyperparameters, and thus degrade markedly when applied to novel domains. We bridge this gap