Masters Thesis: Steering controller identification and design for human-like overtaking

DAVI project, which is the abbreviation for Dutch Automated Vehicle Initiative, aims to explore, improve and demonstrate automated driving on public roads. The ultimate goal of the project is to achieve a full autonomous vehicle without any driver, but only users on it. Compared with traditional human driving, automated driving is an appealing topic due to its advantages of enhancing road safety, increasing road capacity and reinforcing driving comfort. Concentrating on overtaking on a two-lane highway, this particular study investigates driver’s steering behavior and experimentally develop a customized controller. In a typical configuration of driver model, the assumption, that a driver conceives a trajectory before the process of overtaking, is problematic in this case due to inaccurate reflection of driving behavior and computational complexity. As an alternative to this method, a target and control scheme is implemented to mimic human overtaking behavior. First, data from driving simulator experiments are provided by over 40 participants, where each driver performs a series of four consecutive overtaking (FCO). Second, parameters of the steering controller are identified to match the overtaking data exhibited in the driving simulator. Furthermore, the controller analysis is carried out to test the performance. Finally, an operating system designed for customized overtaking is programmed using MATLAB guide. A human-machine interface demonstrates how the system is applied to a real autonomous vehicle. The closedloop simulation result suggests key characteristics of overtaking are properly captured and computational speed of the algorithm is sufficiently fast, thus indicates a solution to practical autonomous overtaking. Master of Science Thesis Yujie Zhang