Seite - (000009) - in Maximum Tire-Road Friction Coefficient Estimation

Abstract The extent to which a certain driver input causes a desired vehicle reaction depends strongly on the road and tire conditions. This also holds true for the interventions of advanced driver assistance systems (ADAS). ADAS are currently developed to meet re- quirements for accident prevention on dry roads. This is necessary in order to avoid incorrect ADAS interventions due to unreliable information. The present work describes a method for estimating the current road and tire conditions based on on-board sensors that measure and process the vehicle’s dynamic state. The first part of this thesis addresses the first research topic, namely determining which of the investigated vehicle state variables are sensitive to road and tire condi- tions, and are therefore theoretically suitable for estimating road and tire conditions. To this end, a sensitivity analysis was conducted by means of a two-track vehicle model described by a non-linear ordinary differential equation system. As expected, the wheel rotational speeds showed the highest sensitivity to the road and tire conditions for all investigated driving states. The second part of the thesis presents a method for esti- mating the road and tire conditions based on the sensor signals present in a vehicle equipped with electronic stability control. To this end, a particle filter is applied, which serves as a non-linear observer within the Bayesian probability framework and makes it possible to solve a state estimation problem while dealing with measurement noise and uncertainties. The research investigated whether the road and tire conditions can be estimated accurately enough to be used within the intervention strategy of an adaptive emergency brake system (AEB). The results show that the proposed vehicle-dynamics-based method has the potential to improve existing techniques for estimating current road and tire conditions. Since the proposed method is based on the measurement of the dynamic reactions of the ve- hicle, the estimation accuracy increases with increasing dynamic excitation. The results demonstrate that the required accuracy for an AEB can be reached for certain driving states.