End-to-end globally consistent registration of multiple point clouds

We present a novel, end-to-end learnable, multiview 3D point cloud registration algorithm. Registration of multiple scans typically follows a two-stage pipeline: the initial pairwise alignment and the globally consistent refinement. The former is often ambiguous due to the low overlap of neighboring point clouds, symmetries and repetitive scene parts. Therefore, the latter global refinement aims at establishing the cyclic consistency across multiple scans and helps in resolving the ambiguous cases. In this paper we propose the first end-to-end algorithm for joint learning of both parts of this two-stage problem. Experimental evaluation on benchmark datasets shows that our approach outperforms stateof-the-art by a significant margin, while being end-to-end trainable and computationally less costly. A more detailed description of the method, further analysis, and ablation studies are provided in the original CVPR 2020 paper [11].