Cooperative Vehicle Guidance Controller (Diplomathesis)


Hanen Soufi

16.10.2013, 14:00, room 4981


The field of highly automated vehicles is rapidly growing worldwide. Such a vehicle typically consists of multiple layers. Localization is done via image processing and GPS, sophisticated navigation is utilized and finally there are controllers for trajectory planning and the low-level dynamics of the vehicle. Although the vehicle is able to operate autonomously, sensing the environment is highly complex and the occurrence of errors cannot be completely excluded. Therefore, the driver needs to be able to overrule the decisions of the vehicle in certain situations. This overruling needs to happen in a smooth and intuitive fashion since both confusion of the driver and sudden movements of the car can lead to severe consequences.

This work investigates the collaboration of a trajectory planner and a human driver. Toward this, driver models were examined, adjusted to an existing vehicle model, and fitted to measurements of a human test driver. Furthermore, a cooperative admittance control strategy inspired by methods from the area of robotics was adjusted to suitably combine the inputs of the driver and the trajectory planner. After adjusting the parameters of the controller using a robust control approach, its performance was demonstrated through extensive simulation and experimental tests.