DeepRING: Learning Roto-translation Invariant Representation for LiDAR based Place Recognition

TL;DR

DeepRING introduces a novel deep learning approach for LiDAR-based place recognition that achieves roto-translation invariance, improving robustness to perspective changes and outperforming existing methods on public datasets.

Contribution

The paper proposes DeepRING, a method that learns roto-translation invariant features from LiDAR scans using sinogram and magnitude spectrum, and formulates place recognition as a one-shot learning problem.

Findings

DeepRING outperforms comparative methods on public datasets.

The approach maintains high discrimination and invariance in representations.

It demonstrates strong generalization across datasets.

Abstract

LiDAR based place recognition is popular for loop closure detection and re-localization. In recent years, deep learning brings improvements to place recognition by learnable feature extraction. However, these methods degenerate when the robot re-visits previous places with large perspective difference. To address the challenge, we propose DeepRING to learn the roto-translation invariant representation from LiDAR scan, so that robot visits the same place with different perspective can have similar representations. There are two keys in DeepRING: the feature is extracted from sinogram, and the feature is aggregated by magnitude spectrum. The two steps keeps the final representation with both discrimination and roto-translation invariance. Moreover, we state the place recognition as a one-shot learning problem with each place being a class, leveraging relation learning to build representation similarity. Substantial experiments are carried out on public datasets, validating the effectiveness of each proposed component, and showing that DeepRING outperforms the comparative methods, especially in dataset level generalization.