State estimation of vehicle handling dynamics using non-linear robust extended adaptive kalman filter

This paper considers translational and rotational state estimation of vehicle handling dynamics using a non-linear robust filtering method. The robust method is needed to compensate the modelling errors which are caused by a combination of many factors such as: parameter uncertainties, model simplification and non-linearities; these inevitably compromise the filter performance. A non-linear Robust Extended Adaptive Kalman Filter (REAKF) is proposed, based on a six degree of freedom (6DOF) vehicle model. The model uncertainties are represented using an integral quadratic constraint method. Parameter adaptation is then included to further increase the accuracy of the filter; the tyre longitudinal and cornering stiffnesses are varied, since both of these parameters are central to the vehicle non-linearities. The filter is tested in both simulation and practice using data collected from a real vehicle. Comparisons are also made with the standard Extended Kalman Filter (EKF) and the non-adaptive Robust Extended Kalman Filter (REKF). The combination of evidence from both tests illustrates the performance and stability benefits of the REAKF.