Hybrid Kinematic-Dynamic Sideslip and Friction Estimation

Abstract

Vehicle sideslip and tyre/road friction are crucial variables for advanced vehicle stability control systems. Estimation is required since direct measurement through sensors is costly and unreliable. In this paper, we develop and validate a sideslip estimator robust to unknown road grip conditions. Particularly, the paper addresses the problem of rapid tyre/road friction adaptation when sudden road condition variations happen. The algorithm is based on a hybrid kinematic-dynamic closed-loop observer augmented with a tyre/road friction classifier that reinitializes the states of the estimator when a change of friction is detected. Extensive experiments on a four wheel drive electric vehicle carried out on different roads quantitatively validate the approach. The architecture guarantees accurate estimation on dry and wet asphalt and snow terrain with a maximum sideslip estimation error lower than 1.5 deg. The classifier correctly recognizes 87% of the friction changes; wrongly classifies 2% of the friction changes while it is unable to detect the change in 11% of the cases. The missed detections are due to the fact that the algorithm requires a certain level of vehicle excitation to detect a change of friction. The average classification time is 1.6 s. The tests also indicate the advantages of the friction classifiers on the sideslip estimation error.