Comparison of Unscented Kalman Filter in general and additive formulation for state estimation in vehicle dynamics

Advanced driver assistance systems in modern vehicles have gained interest in the past decades. For most of these systems accurate knowledge about the current driving state, describing the vehicle's stability is necessary. Especially information about the vehicle's sideslip angle and rate can improve the performance of many assistance systems, e.g. Electronic Stability Program (ESP). Since the sideslip angle can not be measured directly or only under high cost, online estimation is proposed. Based on measurements taken on a Volkswagen Golf GTE Plug-In Hybrid assumptions for the process noise are made. Considering these noise characteristics in simulation using a detailed nonlinear vehicle dynamic model, the Unscented Kalman Filter in general and additive form is implemented for sideslip angle estimation. Their performance is compared considering various maneuvers.