On the descriptive power of LiDAR intensity images for segment-based loop closing in 3-D SLAM

TL;DR

This paper introduces a deep learning-based method to enhance LiDAR SLAM by generating descriptive location signatures from intensity images, improving loop closure detection through context-aware descriptors.

Contribution

It presents a novel neural network architecture that learns visual context from synthetic LiDAR intensity images to produce more effective descriptors for SLAM.

Findings

Improved descriptiveness of location signatures.

Enhanced reliability of loop closure detection.

Validation on two public datasets.

Abstract

We propose an extension to the segment-based global localization method for LiDAR SLAM using descriptors learned considering the visual context of the segments. A new architecture of the deep neural network is presented that learns the visual context acquired from synthetic LiDAR intensity images. This approach allows a single multi-beam LiDAR to produce rich and highly descriptive location signatures. The method is tested on two public datasets, demonstrating an improved descriptiveness of the new descriptors, and more reliable loop closure detection in SLAM. Attention analysis of the network is used to show the importance of focusing on the broader context rather than only on the 3-D segment.