The previous research on the application of Model Predictive Control (MPC) to aspects of driving a vehicle is considered and extended. The control algorithm differs significantly from previous work and complements previous studies by the application of MPC in the nonlinear domain. A virtual rider for the guidance of a nonlinear vehicle model has been created. The control algorithm is built considering two main loops: one for the real world and one for the mental world in which the driver predicts the future states of the vehicle. Due to its formulation this method can be called Time-Variant MPC (TV-MPC), and it is loosely comparable with a multi-model MPC approach, but the solution is not dependent on the displacement about a trim state but only on the time step defined in the integration settings. The TV-MPC driver capability is assessed by U-turn manoeuvres and considerations on the effective preview distance and on the information used by the driver in order to accomplish the task are included.