Real-time LIDAR localization in natural and urban environments

TL;DR

This paper presents a real-time LIDAR localization method using deep learning for efficient point cloud processing, enabling deployment on resource-constrained robots across diverse environments.

Contribution

Develops a fast, memory-efficient LIDAR localization approach with a novel deep learning architecture suitable for online use on limited hardware.

Findings

Achieves threefold reduction in computation time

Reduces memory usage by 70%

Maintains localization accuracy across diverse environments

Abstract

Localization is a key challenge in many robotics applications. In this work we explore LIDAR-based global localization in both urban and natural environments and develop a method suitable for online application. Our approach leverages efficient deep learning architecture capable of learning compact point cloud descriptors directly from 3D data. The method uses an efficient feature space representation of a set of segmented point clouds to match between the current scene and the prior map. We show that down-sampling in the inner layers of the network can significantly reduce computation time without sacrificing performance. We present substantial evaluation of LIDAR-based global localization methods on nine scenarios from six datasets varying between urban, park, forest, and industrial environments. Part of which includes post-processed data from 30 sequences of the Oxford RobotCar dataset, which we make publicly available. Our experiments demonstrate a factor of three reduction of computation, 70% lower memory consumption with marginal loss in localization frequency. The proposed method allows the full pipeline to run on robots with limited computation payload such as drones, quadrupeds, and UGVs as it does not require a GPU at run time.