DRO: Doppler-Aware Direct Radar Odometry

TL;DR

DRO tackles robust ego-motion estimation for spinning mmwave radars by formulating a direct SE(2) odometry framework that uses all radar intensity data without extracting features. It jointly optimizes a direct intensity term and a Doppler-based constraint, with an on-the-fly local map and GP-based Doppler infill to handle motion distortion and velocity observability under triangular modulation. The approach, aided by a gyroscope, achieves state-of-the-art results on automotive datasets (Boreas/MulRan) and demonstrates strong performance in off-road scenarios, with real-time GPU implementation and publicly available code. This work significantly enhances radar-based navigation under adverse weather and feature-deprived environments, enabling robust odometry where cameras or lidars may fail. It also provides a flexible, training-free framework that can adapt to different radar modulation patterns and motion models, paving the way for broader adoption of direct radar odometry in robotics.

Abstract

A renaissance in radar-based sensing for mobile robotic applications is underway. Compared to cameras or lidars, millimetre-wave radars have the ability to `see' through thin walls, vegetation, and adversarial weather conditions such as heavy rain, fog, snow, and dust. In this paper, we propose a novel SE(2) odometry approach for spinning frequency-modulated continuous-wave radars. Our method performs scan-to-local-map registration of the incoming radar data in a direct manner using all the radar intensity information without the need for feature or point cloud extraction. The method performs locally continuous trajectory estimation and accounts for both motion and Doppler distortion of the radar scans. If the radar possesses a specific frequency modulation pattern that makes radial Doppler velocities observable, an additional Doppler-based constraint is formulated to improve the velocity estimate and enable odometry in geometrically feature-deprived scenarios (e.g., featureless tunnels). Our method has been validated on over 250km of on-road data sourced from public datasets (Boreas and MulRan) and collected using our automotive platform. With the aid of a gyroscope, it outperforms state-of-the-art methods and achieves an average relative translation error of 0.26% on the Boreas leaderboard. When using data with the appropriate Doppler-enabling frequency modulation pattern, the translation error is reduced to 0.18% in similar environments. We also benchmarked our algorithm using 1.5 hours of data collected with a mobile robot in off-road environments with various levels of structure to demonstrate its versatility. Our real-time implementation is publicly available: https://github.com/utiasASRL/dro.

Figures (10)

• Figure 1: This paper presents a Direct Radar Odometry (DRO) method that provides state-of-the-art performance in a wide range of environments, including unstructured ones (right) and geometrically degenerated ones such as featureless tunnels (left).
• Figure 2: Impact of different frequency modulation patterns (top) and Doppler effect on the data collected with spinning fmcw radars (bottom).
• Figure 3: Overview diagram of the proposed direct radar odometry method. To leverage the Doppler-based velocity constraint, the radar must use a triangular frequency modulation pattern. Note that ${O_{{\psi_{}}}}$ accounts for motion and Doppler-induced distortion.
• Figure 4: Illustration of the proposed infilling. The raw Doppler distorted radar data (left) is split between up and down-chirp azimuths. Gaussian Process (GP) regression is used to interpolate every second azimuth (middle two, the interpolated portion is in grey). Both ${\mathcal{I}^{\uparrow}}$ and ${\mathcal{I}^{\downarrow}}$ virtually observe the same geometry without assuming that two consecutive azimuths of the raw data observe the same part of the environment (right, with blue positive and red negative).
• Figure 5: Illustration of the GP-based infill highlighting the Doppler effect on radar data collected with a triangular frequency modulation pattern (the blue arrow is the sensor velocity).