Seite - 88 - in Integration of Advanced Driver Assistance Systems on Full-Vehicle Level - Parametrization of an Adaptive Cruise Control System Based on Test Drives

6. Upper Level Controller Parameter Identification vehicle was positioned with its initial speed from the measurement at the measured dis- tance behind the OTF. The desired speed is set to vset = 100km/h, and the desired time gap to τset= 1.2s, as recorded in the measurement. Figure 6.13 shows the compar- ison between the measurement and the simulation for the longitudinal vehicle speed vvx, the longitudinal vehicle acceleration vax, and the inter-vehicle distance srOTF and range rate sr˙OTF andTTCOTF. In the simulation, theACCvehiclebegins toaccelerate earlier than in reality. As a result, with the same level of maximum acceleration, the vehicle in reality has an overshoot in speed and in the inter-vehicle distance signals. That is not the case in the simulation. Due to the small levels of acceleration and deceleration of the OTF, theovershoot in the realmeasureddatamaybecritical. Dangerous situationsmay occur if the OTF performs a hard deceleration manoeuvre. The delay for the accelera- tion manoeuvre is not critical because it will never lead to a critical situation. For both manoeuvres, the difference in time is nearly the same at about 3s. The measured ACC vehicle may suppress the deceleration command until a certain TTC is reached. At the measured vehicle, it begins to decelerate at TTCOTF ≈ 10s. Fancher et al. described in [FBE01] that drivers become anxious if the TTC falls below 9 to 10s. Thus, the measured ACC vehicle is near the given boundary, and some drivers may feel scared. In comparison, the simulated ACC system begins to decelerate even at small inter-vehicle velocities, resulting in a TTCOTF ≈ 69s, see fig. 6.13. If it is necessary to delay the beginning of the deceleration, this could easily be implemented in the controller. The desired acceleration of the ACC controller should only be suppressed until the first time TTC falls below the defined limit. Since simulations cannot answer questions about which of the two behaviours drivers prefer, simulator tests or real test drives have to be performed with both settings and a high number of probands. This will lead to the problem that even if most of the probands prefer one setting, there will still be drivers who prefer the other setting. A compromise for dealing with this problem is that both settings are made available in the vehicle, which has already been implemented in production vehicles. Drivers must select their preferred setting via the Human-Machine Interface (HMI). To sum up, the identified parameters meet all the requirements. String stability is guaranteed, the motion of the ego vehicle in the measurements is nearly duplicated by the ACC-equipped vehicle, and the performance of the system is similar to a production vehicle equipped with an ACC system. The identification itself is fast, if the required measurementsareavailable. Thebigadvantage is that if themeasurementshavealready beenmade, thedata canbeused toparametrize controller typesother than theoneused in this work. In the future, the identified parameters should be evaluated on a driving simulator. 88