To meet the needs of comfort and take the vehicle safety into account at the same time of car following model of the adaptive cruise control (ACC) system, a multi-objective robust control algorithm of car following model is designed based on model predictive control (MPC) theory. A mutually longitudinal kinematics model considering the pre-car acceleration noise is presented to fully reflect the dynamic evolution of ACC system and improve the model's accuracy and reliability. The target of the requirement of ACC system is analyzed and a multi-objective MPC algorithm considering comfort and safety is proposed. By introducing a correction term feedback, the robustness of the control system is improved for the reason of the MPC algorithm's sensitive to disturbance. By adopting vector management method, the problem is solved which the MPC algorithm can't find the optimum solution caused by hard constraints. The simulation shows that the designed control algorithm ensures that the acceleration and jerk of the ACC system is kept in the comfort range, while the distance between the pre-vehicle and host-vehicle is always greater than the minimum safe distance.