Uncertainty Quantification for Radar/Inertial Pose Estimation in GNSS-Denied Areas

In this paper, we first design and analyze a high-fidelity automotive radar radial velocity measurement model for vehicle velocity estimation. We then use this model in a tightly-coupled extended Kalman filter (EKF) algorithm for tight radar/IMU integration (IMU stands for inertial measurement unit). We use both analytical covariance derivation and numerical simulations to quantify the reduction in velocity and positioning error drift of the radar/IMU system as compared to a free-coasting IMU.