摘要：AbstractWithin the scope of highly automated driving, steering a vehicle autonomously along a pre-specified geometric reference path is an important control task. In this article we present an implementation of model predictive path-following control for a modified VW Golf VII series vehicle. This approach allows the assignment of the reference motion and the computation of inputs to track the reference simultaneously at the runtime of the controller, while respecting actuator and state constraints. In contrast to trajectory tracking, the time evolution along the path is not fixed a priori. It can be seen as a degree of freedom to be determined by the controller. This allows the controller to adapt the vehicle speed to the respective driving situation according to the modeled physical limits of the vehicle. Therefore, a combined longitudinal and lateral vehicle dynamic model is derived firstly. Finally, experimental results are presented for a test scenario to demonstrate the abilities of the approach.
关键词：Keywordsautonomous vehiclesconstraintsdynamic modellingnonlinear model predictive controloptimal controlpath followingreal-time controlvehicle dynamics
