SE3ET: SE(3)-Equivariant Transformer for Low-Overlap Point Cloud Registration

TL;DR

SE3ET introduces an SE(3)-equivariant transformer framework that enhances robustness and accuracy in low-overlap, large-transformation point cloud registration for robotics applications.

Contribution

The paper presents SE3ET, a novel equivariant learning framework using point convolution and transformers for improved 3D point cloud registration.

Findings

Effective on indoor and outdoor benchmarks with arbitrary transformations.

Robust performance under low-overlap conditions.

Competitive run-time performance.

Abstract

Partial point cloud registration is a challenging problem in robotics, especially when the robot undergoes a large transformation, causing a significant initial pose error and a low overlap between measurements. This work proposes exploiting equivariant learning from 3D point clouds to improve registration robustness. We propose SE3ET, an SE(3)-equivariant registration framework that employs equivariant point convolution and equivariant transformer designs to learn expressive and robust geometric features. We tested the proposed registration method on indoor and outdoor benchmarks where the point clouds are under arbitrary transformations and low overlapping ratios. We also provide generalization tests and run-time performance.