DRIO: Robust Radar-Inertial Odometry in Dynamic Environments

Accurate and robust localization is essential for mobile robots. Recently, millimeter wave (mmWave) radars have been widely used for odometry, owing to their robustness to all-weather conditions, lightweight and low cost. However, existing radar-based odometry methods degrade severely in high-dynamic environments. In this letter, we propose a robust radar-inertial odometry method for high-dynamic environments (DRIO) by exploiting the ground, an ever-present static target that is unaffected by the dynamic environments. The points of the ground surface were traditionally treated as clutter points in previous works due to their unstable distribution. We overcome this limitation by detecting ground points using both Doppler and geometric characteristics. During the detection process, accurate radar velocity is jointly estimated, which is then fused with inertial data to obtain the odometry. The real-world evaluations indicate that the proposed method achieves robust and Lidar-level localization in complex dynamic environments. In addition to odometry, our method can effectively improve the quality of radar point clouds for subsequent perception tasks.