A quality-aware robust extended Kalman filter for low-cost GNSS/IMU vehicle localization in urban canyons

Abstract

Accurate vehicle localization in urban canyons using low-cost Global Navigation Satellite System (GNSS) and Inertial Measurement Unit (IMU) is challenging due to multipath and non-line-of-sight (NLOS) receptions. Conventional loosely coupled Extended Kalman Filters (EKFs) with fixed covariance often over-rely on degraded GNSS updates. This paper proposes a lightweight, quality-aware robust EKF. The method incorporates three practical mechanisms: a GNSS horizontal-accuracy-driven dynamic measurement covariance, a Normalized Innovation Squared (NIS)-based adaptive scaling rule, and a scheduled GNSS update strategy. Experiments on the challenging SmartLoc Berlin Potsdamer Platz sequence show that the proposed method achieves a position Root Mean Square Error (RMSE) of 5.761 m, outperforming the standard EKF (5.856 m) and raw GNSS (5.892 m). The yaw RMSE is reduced to 4.856°, with 11 adaptive updates triggered along the trajectory. The method is model-compact, fully reproducible, and achieves measurable gains without requiring raw pseudorange processing, maps, or vision sensors. This provides an engineering-oriented solution for low-cost urban vehicle localization.