Page - 7 - in Maximum Tire-Road Friction Coefficient Estimation

1 Introduction WarningADASnotifythedriver,butdonotdirectly influencethemotionofthevehicle like intervening ADAS do. Especially on the course planning level, information about the environment (e.g. other traffic participants, road geometry) is required to identify possible critical situations in advance. This leaves more time for a plausibility check of estimates, if thesearesubject tocontrol inputs forADAS.Nevertheless, therequirements for an estimate of the friction potential µmax depend on the specific application at the vehicle level. In the following section, the potential ADAS and VDC that could be improved by adapting their strategies to the current friction potential are identified first, see Section 1.3.1. The next sections then discuss the specific requirements for an online estimateof the frictionpotential tobeapplied in specificADAS, seeSection1.3.2, and in specific VDC, see Section 1.3.3. The requirements cover the tolerable time delay τM before a robust estimate is available and the tolerable deviation ∆µ. The tolerable time delay is a theoretical value and considers the computational time of the friction potential in general, but also for changing states of either the road condition or the vehicle’s state (e.g. from free rolling to accelerating). The relation between an estimate µ̂ and an ideal valueµref is given by µ̂=µref±∆µ. (1.1) 1.3.1. Significance of friction potential for ADAS and VDC Eichberger, Tomasch et al. evaluated the potential benefit of specific ADAS and VDC for preventing or mitigating accidents in their RCS-TUG Study, [ETR+10]. They per- formed a case-by-case analysis within database for 514 incidents of all 848 accidents in Austria in 2003 in which people died. The basis of this investigation was ZEDATU2, a central database in Austria for in-depth analysis of road accidents with a focus on fatal car accidents, [TS06]. Experts reconstructed the collision phase using the commercial software PC Crash, [SM01]. With the data from the collision phase, numerical recon- structionof thepre-collisionphasewaspossible. Altogether, theeffectof 43warningand intervening systems on the simulated real life scenarios were investigated. The benefit of each system was determined using both the number of accidents prevented and the cases where fatalities were avoided with a high probability, see Eichberger for a detailed explanation, [ETR+10, p.83-97]. Figure 1.5 shows the top ten systems identified in Eichberger et al. for avoiding or 2Abbreviation in German forZentraleDatenbank zurTiefenanalyse von Verkehrsunfa¨llen 7