The limitations of brain-computer interfaces (BCI) lead to a poor control performance of brain-controlled assistive vehicles through a BCI. To improve the control performance and safety of brain-controlled assistive vehicles, this paper proposes a novel assistant controller designed by using the model predictive control method. The proposed controller combines the tracking of the driver's intention and safety of the vehicle into an optimization problem so as to make the driver be in charge of the vehicle as much as possible when the vehicle is safe. The simulation result shows that the assistive system can improve the control performance and safety of the brain-controlled vehicles.