Dimensionless sensitivity methods to identify vehicle cornering stiffness from yaw rate measurements

A significant amount of research has focused on model-based identification of vehicle behavior using Kalman-filter or similar approaches with sometimes complex, high-order or nonlinear vehicle models to achieve estimation accuracy. This work examines the model complexity versus accuracy tradeoff with a bias toward greatly reducing the complexity of the identification model even if this allows some identification inaccuracy. By using the simplest model possible, but no simpler, the goal is to achieve fast convergence. Model simplification is obtained using a novel dimensionless method that exposes explicit and implicit coupling between Bode parameter sensitivities, a coupling that constrains the possible parameter variations. To demonstrate this method, vehicle yaw rate data is used to attempt to identify the cornering stiffness parameter governing the tire-road interaction. Simulation results and experimental implementation on a research vehicle under changing road conditions are presented.