Unsupervised Point Cloud Registration with Self-Distillation
This addresses the scalability issue in training deep learning models for point cloud registration in robotics and autonomous driving by eliminating the need for costly ground truth data, though it is incremental as it builds on existing self-distillation and robust solver concepts.
The paper tackles the problem of rigid point cloud registration without ground truth poses by introducing a self-distillation approach that uses a teacher network with a robust solver to optimize for unsupervised inlier ratio, achieving state-of-the-art results on the 3DMatch benchmark and demonstrating generalization to automotive radar data.
Rigid point cloud registration is a fundamental problem and highly relevant in robotics and autonomous driving. Nowadays deep learning methods can be trained to match a pair of point clouds, given the transformation between them. However, this training is often not scalable due to the high cost of collecting ground truth poses. Therefore, we present a self-distillation approach to learn point cloud registration in an unsupervised fashion. Here, each sample is passed to a teacher network and an augmented view is passed to a student network. The teacher includes a trainable feature extractor and a learning-free robust solver such as RANSAC. The solver forces consistency among correspondences and optimizes for the unsupervised inlier ratio, eliminating the need for ground truth labels. Our approach simplifies the training procedure by removing the need for initial hand-crafted features or consecutive point cloud frames as seen in related methods. We show that our method not only surpasses them on the RGB-D benchmark 3DMatch but also generalizes well to automotive radar, where classical features adopted by others fail. The code is available at https://github.com/boschresearch/direg .