RAVE: A Framework for Radar Ego-Velocity Estimation

TL;DR

RAVE addresses radar ego-velocity estimation by integrating zero-velocity detection, robust inlier-based velocity estimation, and a simple feasibility filter to reject implausible results. It analyzes multiple outlier rejection strategies and robust loss functions, validating the approach on IRS, ColoRadar, and VoD datasets and demonstrating improved odometry when integrated with MOVRO. The contributions include a modular, open-source Radar Velocity Estimator with configurable loss and rejection options, plus empirical evidence of enhanced estimation and downstream performance in challenging environments.

Abstract

State estimation is an essential component of autonomous systems, usually relying on sensor fusion that integrates data from cameras, LiDARs and IMUs. Recently, radars have shown the potential to improve the accuracy and robustness of state estimation and perception, especially in challenging environmental conditions such as adverse weather and low-light scenarios. In this paper, we present a framework for ego-velocity estimation, which we call RAVE, that relies on 3D automotive radar data and encompasses zero velocity detection, outlier rejection, and velocity estimation. In addition, we propose a simple filtering method to discard infeasible ego-velocity estimates. We also conduct a systematic analysis of how different existing outlier rejection techniques and optimization loss functions impact estimation accuracy. Our evaluation on three open-source datasets demonstrates the effectiveness of the proposed filter and a significant positive impact of RAVE on the odometry accuracy. Furthermore, we release an open-source implementation of the proposed framework for radar ego-velocity estimation accompanied with a ROS interface.

Figures (5)

• Figure 1: The ego velocity model relies on data from stationary targets (shown in green). Radars measure the relative velocity based on the Doppler effect (indicated by the dashed arrow). Outliers, such as dynamic targets or ghost targets, are indicated by red dots.
• Figure 2: The proposed RAVE framework. Through our filtering method we determine the feasibility of the estimated velocity. The accepted values serve as inputs for the next step, while rejected values are discarded.
• Figure 3: Visualization of velocity estimation with and without the proposed filter on the Mocap dark fast sequence.
• Figure 4: Visualization of velocity estimation with three different methods on the aspen_run0 sequence.
• Figure 5: MOVRO movro trajectories on the VoD dataset for four different approaches to radar ego velocity estimation.