Diverse vehicle active safety systems including vehicle electronic stability control (ESC) system, anti-lock braking system (ABS), and traction control system (TCS) are significantly relying on information about the vehicle's states and parameters, as well as the vehicle's surroundings. However, many important states or parameters, such as sideslip angle, tire-road friction coefficient, road gradient and vehicle mass are hard to directly measure, and hence advanced estimation algorithms are needed. Furthermore, enhancements of sensor technologies and the emergence of new concepts such as Internet of Things and their automotive version, Internet of Vehicles, facilitate reliable and resilient estimation of vehicle states and road conditions. Consequently, developing a resilient estimation structure to operate with the available sensor data in commercial vehicles and be flexible enough to incorporate new information in future cars is the main objective of this thesis.

This thesis presents a reliable corner-based vehicle velocity estimation and a road condition classification algorithm. For vehicle velocity estimation, a combination of vehicle kinematics and the LuGre tire model is introduced in the design of a corner-based velocity observer. Moreover, the observability condition for both cases of time-invariant and parameter varying is studied. The effect of suspension compliance on enhancing the accuracy of the vehicle corner velocity estimation is also investigated and the results are verified via several experimental tests.

The performance and the robustness of the proposed corner-based vehicle velocity estimation to model and road condition uncertainties is analyzed. The stability of the observer is discussed, and analytical expressions for the boundedness of the estimation error in the presence of system uncertainties for the case of fixed observer gains are derived. Furthermore, the stability of the observer under arbitrary and stochastic observer gain switching is studied and the performances of the observer for these two switching scenarios are compared. At the end, the sensitivity of the proposed observer to tire parameter variations is analyzed. These analyses are referred to as offline reliability methods.

In addition to the off-line reliability analysis, an online reliability measure of the proposed velocity estimation is introduced, using vehicle kinematic relations. Moreover, methods to distinguish measurement faults from estimation faults are presented. Several experimental results are provided to verify the approach.

An algorithm for identifying (classifying) road friction is proposed in this thesis. The analytical foundation of this algorithm, which is based on vehicle response to lateral excitation, is introduced and its performance is discussed and compared to previous approaches. The sensitivity of this algorithm to vehicle/tire parameter variations is also studied. At the end, various experimental results consisting of several maneuvers on different road conditions are presented to verify the performance of the algorithm.