The paper proposes a combined control design framework using Model Predictive Control (MPC) and ultralocal model-based methods. The main idea behind the control algorithm is to exploit the advantage of both approaches. During the control input computation, a simplified model is used, which has a significant impact on the computational cost. Moreover, the simplified model does not contain hardly measurable or varying vehicle-specific parameters, which makes the whole control design process easier. The ultra-local model is used to deal with the unmodeled dynamics of the vehicle, by which the performance of the control system can be increased. The effectiveness of the proposed control structure is demonstrated through trajectory tracking problem of autonomous vehicles. The whole algorithm is implemented in a high-fidelity vehicle dynamics simulation software, whose results are compared to an accurate model-based MPC in terms of computational cost and tracking accuracy.