An evaluation of CFEAR Radar Odometry

TL;DR

The paper evaluates the CFEAR radar odometry method for spinning 2D radar data, aiming at accurate, real-time localization across varied environments. It presents a real-time-capable configuration using the original parameter set and demonstrates low drift on the Boreas dataset, with an improved implementation enabling tighter robustness and higher frame rates. The authors provide open-source releases of the code and evaluation to facilitate reproducibility. The work contributes a practical, robust radar odometry solution with competitive drift performance in publicly available benchmarks, enabling deployment in real-time robotics.

Abstract

This article describes the method CFEAR Radar odometry, submitted to a competition at the Radar in Robotics workshop, ICRA 20241. CFEAR is an efficient and accurate method for spinning 2D radar odometry that generalizes well across environments. This article presents an overview of the odometry pipeline with new experiments on the public Boreas dataset. We show that a real-time capable configuration of CFEAR - with its original parameter set - yields surprisingly low drift in the Boreas dataset. Additionally, we discuss an improved implementation and solving strategy that enables the most accurate configuration to run in real-time with improved robustness, reaching as low as 0.61% translation drift at a frame rate of 68 Hz. A recent release of the source code is available to the community https://github.com/dan11003/CFEAR_Radarodometry_code_public, and we publish the evaluation from this article on https://github.com/dan11003/cfear_2024_workshop