Sensitivity-based Road Friction Estimation in Vehicle Dynamics using the Unscented Kalman Filter

Automotive research and development passed through an enormous development within the last decades. In particular focusing on the goal of autonomous driving an tremendous potential is expected in the upcoming years. In this regard a robust and exact perception of the vehicle's environment is necessary. Especially, the road condition, represented by the friction coefficient between tires and road, has major influence on the vehicle's behavior and thus its stability. Therefore, robust online estimation of the friction coefficient is indispensable for autonomous driving to ensure save driving. In this paper online estimation of the bounded maximum friction coefficient based on serial sensors is presented using a sensitivity-based joint unscented Kalman filter. To achieve robust estimation without parameter estimation drift during phases of insufficient excitation, a local sensitivity analysis is introduced. The friction estimation results are validated by utilizing measurements taken on a Volkswagen Golf GTE Plug-In Hybrid on four different road surfaces for longitudinal and lateral dynamic maneuvers.