Design, experimental validation, and comparison of two model-based EKF observers for lateral vehicle dynamics estimation

One of the most important tasks of the vehicle dynamics estimation problem is to obtain the longitudinal, lateral, and vertical forces acting upon the wheels. These variables are fundamental for studying vehicle controllability and stability. However, they are difficult to obtain in terms of measurement techniques. Regarding lateral/tire forces, two observers based on the state-observer Extended Kalman Filter (EKF) theory and on the dynamic responses of a vehicle instrumented with standard sensors are proposed, discussed and compared in this study. The first observer estimates the lateral forces per axle using a random walk model, and then calculates the individual lateral tire forces on the basis of the vertical tire forces distribution. The second observer directly estimates the lateral forces per tire using a transient relaxation model. The developed observers are able to work in real-time in normal and in critical driving situations. Performances are tested using a laboratory car running on a wet track.