Page - 90 - in Integration of Advanced Driver Assistance Systems on Full-Vehicle Level - Parametrization of an Adaptive Cruise Control System Based on Test Drives

7. Summary and Conclusion (MPC), fuzzy control and Sliding Mode Control (SMC), each of which showed different behaviour. The best performance is delivered by the CTG controller and the SMC. Chapter 3: Development Process. This chapter described the V-Model, which is the common development process for electric and electronic systems. The development of ADAS is a trade-off between shortening the development and validation time to save costs and delivering a system that satisfies the customer. The main problem is that not all situations that might occur can be tested. This may lead to an infinite number of test cases. To achieve a high number of tests within a short time, Hardware-in-the-Loop (HIL) and Model-in-the-Loop (MIL) tests were used. Although they cannot completely replace expensive, time-consuming real vehicle tests, theycanreduce thenumberof such tests required. Chapter 4: Measurements. Tests with non-professional test drivers and a specially equipped vehicle were carried out. The probands drove a vehicle called the ego vehicle, with a production RADAR sensor mounted on its front. Additionally, an extended vehicle dynamics measurement system was mounted on the ego vehicle and on another vehicle. With this measurement system, the relative motion between the ego vehicle and the other vehicle was measured with an accuracy of a few centimetres. With this measurement setup, abasic andprobandstudywere conducted. Twelvedifferentdrivers travelled an overall distance of 445.4km on a defined route in and around the city of Graz. Sincethetestsweredoneonpublic roads, thesideslipanglecouldnotbemeasured directly. Therefore, a linear observer was created, which delivered satisfying results. Chapter 5: Selection of the Object to Follow. This chapter compared different path prediction and object selection algorithms. This study was based on the mea- surements made in chapter 4. The evaluated path prediction algorithms were constant curvature algorithms andalgorithmsbasedonthe linearSingle-Track Model (STM).The constant curvature algorithms predict the vehicle path with the hypothesis that the ac- tual measured curvature of the path will stay constant in the future. The input for the linear STM was the steering angle, and the output was the predicted side slip angle and the vehicle yaw rate. First, the actual steering angle was set constant for the input of the STM. As a second option, a novel steering angle prediction algorithm was developed, which was used as an input for the STM. These path prediction algorithms were applied to all time steps of the measurement. The predicted paths were compared to the driven paths, using the measurement data recorded during the test drives described in chap- ter4. Thepredictionwasperformedfor twotimehorizons. Forashortpredictiontimeof three seconds, there was hardly any difference between the algorithms. Three seconds is a typical prediction time for safety systems, such as Forward Collision Warning (FCW) and Automatic Emergency Brake (AEB) systems. At the long prediction horizon of ten seconds, there were differences in the evaluation. The best option was the combination of the new steering algorithm and the linear STM. A prediction time of ten seconds is important for ACC systems. The predicted path was used to select the Object to Follow (OTF). There, two different object selection algorithms were compared. The simplest one selects the nearest object 90