Global visual localization in LiDAR-maps through shared 2D-3D embedding space

TL;DR

This paper introduces a novel deep learning method that creates a shared embedding space for images and LiDAR point clouds, enabling effective global visual localization within LiDAR maps, which is a largely unexplored area.

Contribution

The work develops a shared 2D-3D embedding space using deep neural networks for cross-modal place recognition between images and LiDAR maps, advancing localization techniques.

Findings

Effective cross-modal place recognition demonstrated on Oxford Robotcar Dataset.

Robustness across different weather and lighting conditions.

Comparison of various learning paradigms and network architectures.

Abstract

Global localization is an important and widely studied problem for many robotic applications. Place recognition approaches can be exploited to solve this task, e.g., in the autonomous driving field. While most vision-based approaches match an image w.r.t. an image database, global visual localization within LiDAR-maps remains fairly unexplored, even though the path toward high definition 3D maps, produced mainly from LiDARs, is clear. In this work we leverage Deep Neural Network (DNN) approaches to create a shared embedding space between images and LiDAR-maps, allowing for image to 3D-LiDAR place recognition. We trained a 2D and a 3D DNN that create embeddings, respectively from images and from point clouds, that are close to each other whether they refer to the same place. An extensive experimental activity is presented to assess the effectiveness of the approach w.r.t. different learning paradigms, network architectures, and loss functions. All the evaluations have been performed using the Oxford Robotcar Dataset, which encompasses a wide range of weather and light conditions.