Robust Estimation of Road Friction Coefficient for Vehicle Active Safety Systems.

Abstract

Vehicle active safety systems stabilize the vehicle by controlling tire forces. They work well only when the tire force command computed by the safety systems is within the friction limit. Therefore, knowledge of the tire/road friction coefficient is important to improve their performance. The objective of this dissertation is to develop a robust friction coefficient estimation algorithm for vehicle active safety systems. The algorithm should be operational in a wide range of vehicle states, robust to plant uncertainties, and use information from sensors that are readily available on typical passenger vehicles. This study presents two methods of estimating the friction coefficient: a lateral dynamics based method and a longitudinal dynamics based method. These two methods are then integrated to improve the working range and robustness of the estimator. The first method is a nonlinear observer based on vehicle lateral/yaw dynamics and the Brush tire model, whereas the second method is a recursive least squares method based on the relationship between tire longitudinal slip and traction force. The two methods are complementary to each other because they rely on different excitation conditions. Therefore, they can be integrated by a switching method where the switching signal depends on the level and kind of excitation. The performance of the estimation algorithm was verified using simulations and test data under a wide range of friction and speed conditions. The test was performed on three different road surfaces: concrete, snow, and ice. The algorithm is able to estimate the friction coefficient of these three surfaces, including during abrupt surface changes and tracks the friction coefficient variance. It exhibits reasonable performance under various driving conditions based on the basic sensors used in vehicle stability control systems. The overall results from simulations and the experiments demonstrate that the proposed approach has the potential for practical applicability to vehicle active safety control.