Seite - 91 - in Integration of Advanced Driver Assistance Systems on Full-Vehicle Level - Parametrization of an Adaptive Cruise Control System Based on Test Drives

in the predicted path, while the second one assigns a priority to every object in the predicted path. The object with the highest priority is used as the OTF. This priority algorithm seemed to handle the error that arises during the path prediction better than thefirstalgorithm,especiallywhenpassinganothervehicleat lowlateraldistances. Thus, the novel steering prediction algorithm in combination with the linear STM was used to predict the ego vehicle’s path. The OTF was selected by applying the priority algorithm. Chapter 6: Upper Level Controller Parameter Identification. Based on the OTF selected in chapter 5, the parameters for a novel ACC algorithm based on the CTG controller were identified. The proposed parametrization of this controller led to the problem that if it was a string stable parametrization, the comfort did not satisfy the driver, and vice versa. Therefore, a new algorithm was developed to deal with this problem. The parameter identification was performed in three steps. First the stopping distance behind an OTF was identified. The resulting constant correlates very well with the liter- ature. As a second step, the time gap at which the driver was following the proceeding vehicle was identified. To this end, the appropriate parts of the measurements have to be selected. This is done with the inverse of the Time to Collision (TTC), which is an indicator for constant following manoeuvres. As a third and final part, the four pa- rameters needed to parametrize the controller were identified. To this end, the Simplex Method of Nelder-Mead was modified to deal with limited solutions for the parameters. The identified parameters were verified in simulation. First, string stability was proven. Due to the non-linear terms in the controller, it could not be proven analytically. There- fore, simulations were performed, and string stability was guaranteed. The performance of the controller was also checked by simulating the manoeuvres recorded with the probands. The comparison was done between the data recorded with a proband driver and the simulation with a simplified longitudinal vehicle model that was controlled by the ACC controller. The recorded and simulated data showed acceptable similarity. As a last step, simulations were compared to measurements made with a production vehicle equipped with an ACC system. The movement of the OTF was fed into a commercially available simulation tool. These simulations showed that the shape of the simulated movement was similar to the measured one. The only difference was that the production ACCvehiclebegantodecelerateandaccelerate later thanthenewlydevelopedcontroller. The aim of this comparison was not to achieve exactly the same behaviour, but rather to have similar signal shapes. In addition, the comparison showed that the real ACC ve- hicle had an undershoot in the inter-vehicle distance, while the simulated controller had none. The simulation cannot determine which of the two controllers people preferred. Therefore, simulator tests or real test drives have to be performed. Final statement: The presented work provides a well-grounded analysis of existing ACCsystems. Theevaluationmethodsdescribedcouldbeusedto furtherdevelopADAS algorithms. One example for an improvement is the novel steering angle prediction algorithm, which showed the best results in combination with the linear STM in the evaluation of the path prediction algorithms. It is possible to use the predicted steering 91