To tackle path tracking accuracy and stability degradation due to steering delay and curvature variation, a model predictive control strategy with delay compensation and curvature augmentation has been developed. Firstly, a first-order inertial model is established to represent the response delay of the steering system caused by factors such as signal transmission and mechanical friction. Real vehicle steering angle data is collected and the parameters of the steering delay model are identified using the particle swarm optimization algorithm. On this basis, the steering delay model is integrated into the vehicle path tracking error equation to enhance the accuracy of the control system model, and curvature is used as an augmentation term in model predictive control to reduce the disturbance of curvature on tracking accuracy. Finally, the designed control strategy is tested through simulation and real vehicle experiments. Simulation results show that the designed control strategy reduces the lateral deviation and yaw angle deviation by 63.1% and 7.5%, respectively, in high curvature turning conditions, and by 30.9% and 43.4%, respectively, in double lane change conditions, avoiding the steering angle oscillation caused by steering delay, thereby improving the vehicle's path tracking accuracy and stability. Real vehicle tests reveal control strategy reduces lateral deviation and yaw angle deviation by 36.6% and 30.4%, respectively, in double lane change conditions, confirming the effectiveness of the proposed control strategy.