Page - 73 - in Integration of Advanced Driver Assistance Systems on Full-Vehicle Level - Parametrization of an Adaptive Cruise Control System Based on Test Drives

6.2. ACC Controller Parameter Identification using P3 = 0.25 and P4 = 0.2 for small errors to create a comfort-oriented system. However, these parameters cannot satisfy either eq. (6.7) or eq. (6.8), and therefore this set of parameters is not string stable. This could be proven by the simulations of fig. 6.2(b). The applied disturbance of vehicle 1 results in very high accelerations of the proceeding vehicles. At vehicle 18, the inter-vehicle distance sr reaches nearly zero, meaning a very dangerous situation occurs between vehicle 17 and 18. At vehicle 19, the inter-vehicle distance reaches zero, meaning vehicle 19 crashes into 18. This situation shows why a non-string-stable ACC controller is dangerous. To handle this trade-off between comfort and safety, the control law of eq. (6.3) is extended with another term. The resulting control law for the upper level controller reads ades=P1 sinh[P2(er˙+P4er)]+P3 (er˙+P4er), (6.9) which is used for the further investigations in this work. To meet the string stability and comfort requirements, the controller should output very small desired accelerations at small errors and high accelerations at high errors. These requirements lead to the extension with the trigonometric function. Winner et al. used a similar approach in [WDS09]. Figure 6.3 shows the comparison between the segmented controller of Winner et al. and the control law described in eq. (6.9). For both functions, the argument esyn=er˙+P4er was used. 6.2. ACC Controller Parameter Identification To identify the parameter of the ACC controller, the scenarios have to be extracted from the measurements described in chapter 4. Therefore, the condition is defined that the index of the Object to Follow (OTF) should not change for a minimum time of ten seconds. Additionally, the probability of existance (see chapter 4 for a description of the quantity) of the selected object must satisfy the condition pex≥ 99%. This leads to a list of 505 scenarios with a minimum length of Tmin= 10s, a maximum length of Tmax= 253.5s, a mean length of T¯ = 45.4s, and a standard deviation of σT = 39.53s. Figure 6.4 shows the steps of the parameter identification. The selected scenarios are divided into two main groups. The first group is the standstill situation, when the ego velocityandthevelocityoftheOTFequalzero, vvx= 0,and svOTF = 0,seechapter6.2.1. The output of these manoeuvres is the inter-vehicle standstill distance s0. The second group of scenarios are the driving manoeuvres when vvx ≥ 1.5m/s. If the condition 1 TTCOTF ≤0.05s−1 is also satisfied, it is called a constant following manoeuvre, see also chapter 6.2.2. The output of these scenarios is the selected time gap τset. With the rest of the scenarios, the so-called dynamic following manoeuvres, the controller parameters P1 toP4 of theACCcontrollerof eq. (6.9)are identified. The followingchaptersdescribe the steps of the identification in detail. 73