APR: Online Distant Point Cloud Registration Through Aggregated Point Cloud Reconstruction
This work addresses a critical issue for driving safety applications by improving online distant point cloud registration, though it is incremental as it builds on existing autoencoder designs.
The paper tackles the problem of registering LiDAR point clouds from distant moving vehicles, which is challenging due to differing point densities and perspectives, by proposing a novel feature extraction framework called APR that uses an autoencoder to reconstruct aggregated point clouds, resulting in a 7.1% increase in average registration recall on LoKITTI and 4.6% on LoNuScenes compared to state-of-the-art methods.
For many driving safety applications, it is of great importance to accurately register LiDAR point clouds generated on distant moving vehicles. However, such point clouds have extremely different point density and sensor perspective on the same object, making registration on such point clouds very hard. In this paper, we propose a novel feature extraction framework, called APR, for online distant point cloud registration. Specifically, APR leverages an autoencoder design, where the autoencoder reconstructs a denser aggregated point cloud with several frames instead of the original single input point cloud. Our design forces the encoder to extract features with rich local geometry information based on one single input point cloud. Such features are then used for online distant point cloud registration. We conduct extensive experiments against state-of-the-art (SOTA) feature extractors on KITTI and nuScenes datasets. Results show that APR outperforms all other extractors by a large margin, increasing average registration recall of SOTA extractors by 7.1% on LoKITTI and 4.6% on LoNuScenes. Code is available at https://github.com/liuQuan98/APR.