RadarLoc: Learning to Relocalize in FMCW Radar

TL;DR

RadarLoc introduces a deep learning approach with self-attention for accurate 6-DoF global pose estimation from FMCW radar scans, outperforming existing radar and camera relocalization methods on outdoor datasets.

Contribution

The paper presents RadarLoc, a novel end-to-end neural network with self-attention for radar-based relocalization, incorporating geometric constraints to enhance accuracy.

Findings

Outperforms existing radar localization methods

Surpasses deep camera relocalization accuracy

Validated on challenging outdoor datasets

Abstract

Relocalization is a fundamental task in the field of robotics and computer vision. There is considerable work in the field of deep camera relocalization, which directly estimates poses from raw images. However, learning-based methods have not yet been applied to the radar sensory data. In this work, we investigate how to exploit deep learning to predict global poses from Emerging Frequency-Modulated Continuous Wave (FMCW) radar scans. Specifically, we propose a novel end-to-end neural network with self-attention, termed RadarLoc, which is able to estimate 6-DoF global poses directly. We also propose to improve the localization performance by utilizing geometric constraints between radar scans. We validate our approach on the recently released challenging outdoor dataset Oxford Radar RobotCar. Comprehensive experiments demonstrate that the proposed method outperforms radar-based localization and deep camera relocalization methods by a significant margin.