To improve the tracking performance of intelligent vehicle, a lateral controller based on LQR theory and a longitudinal controller based on sliding mode theory were proposed to meet the requirements of tracking accuracy and stability in this paper. Firstly, a feedforward LQR controller was established based on 2-DOF (two degree of freedom) dynamics model. To solve the problem of feedforward LQR controller stability reduce caused by model linearization, a real-time velocity-road curvature fuzzy adaptive prediction LQR controller combined with CRTV model was established. In addition, to improve the stability and tracking accuracy of longitudinal velocity, a longitudinal tracking controller was proposed based on SMC (sliding mode control) theory. To verify the proposed controller, co-simulation and HIL (hardware in loop) experiment were conducted. The results show that the proposed controller combines tracking accuracy and stability. The tracking performance is significantly improved.