A Novel Tuning Approach of the H∞ Filter for Longitudinal Tracking of Automated Vehicles

This paper contributes to the challenging field of reliable vehicle tracking as a pivotal part of automated driving. The Kalman filter is an optimal state estimator based on the assumption of an accurate dynamical model and known noise statistic. In order to achieve a fast, robust and efficient vehicle state estimation in the omnipresence of model imperfections and measurement noise, a synergetic combination of the Kalman filter and H∞ filter is proposed, making optimal usage of their advantages. The performance of the filters and their combination is analyzed throughout a sensor data fusion example. Based on the determination of the position and velocity of a vehicle, the improvements, but also the limits of the different approaches are discussed. Additionally, the possibility to detect sensor faults such as time-varying offsets by augmenting the state-space model is explained.