Seite - 31 - in Maximum Tire-Road Friction Coefficient Estimation

2 Estimation of the friction potential Identification of Friction Potential effect-based cause-based intermediary layers roughness tire tread acoustic e.g. slip-based accelerating braking cornering vehicle-dynamics- based tire-related Figure 2.10.: Classification of the vehicle-fixed methods for determining friction poten- tial, based on Uchanski, [Uch01, p.18]. • Presence of water, snow and ice on the road •Water depth • Precipitation and precipitation density • Road roughness and texture Basedonthesensor information, thecorresponding frictionpotentialhas tobederived using additional relations or models. Among other methods, artificial neural networks (ANN) are used to relate sensor information and friction potentials, e.g. [ISAA10]. ANN, which need to be trained with measurement data, are not able to extrapolate conditions that are not within their training data. Past European research projects that have worked extensively with these methods include the aforementioned programmes PROMETHEUS, e.g. [BBES94], and FRICTI@N, [KP09, p.10-11]. To this point, there has been no single sensor available that can estimate the friction potential continuously in a moving vehicle, [KP09, p.12]. In addition, robustness over the whole life-cycle as well asadditional costsandmaintenancehave tobe taken intoaccountwhenconsidering the use of additional sensors to identify the friction potential. 2.2.2. Effect-based approaches Rather than measuring causes, effect-based approaches observe parameters that are af- fectedbythe frictionpotential. Thesemethodscanbeclassifiedasmethods thatobserve 31