Unsupervised Point Cloud Registration with Self-Distillation

TL;DR

This paper introduces an unsupervised self-distillation method for rigid point cloud registration that leverages a teacher-student network with a robust solver, eliminating the need for ground truth labels and outperforming existing methods.

Contribution

The paper proposes a novel unsupervised learning framework using self-distillation and a robust solver for point cloud registration, removing the reliance on ground truth poses.

Findings

Outperforms existing methods on the 3DMatch benchmark

Generalizes effectively to automotive radar data

Simplifies training by removing handcrafted features and frame dependencies

Abstract

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 .