Adaptive EKF-Based Vehicle State Estimation With Online Assessment of Local Observability

摘要

In this paper, an extended Kalman filter-based estimator adopting a dynamic vehicle model for determining the vehicle’s longitudinal and lateral velocity as well as the yaw rate is proposed. Two additional adaptation states are introduced to scale longitudinal and lateral tire forces if necessary to account for uncertainties in the tire/road contact. As excitation plays a vital role as far as observability is concerned, the suggested approach assesses local observability online and keeps an unobservable adaptation state constant by introducing the respective state as a virtual measurement variable when losing local observability. Furthermore, the filter is part of a Global Navigation Satellite System (GNSS)-based estimation framework. It exploits the availability of a GNSS-based horizontal velocity estimate instead of wheel speeds as aiding measurement, thus being independent of wheel slip. Experimental results for scenarios with different kinds of excitation show the effectiveness of the proposed estimator in the nominal as well as in the perturbed vehicle parameter case requiring filter adaptation.