In order to reject the effects of disturbances on the path tracking accuracy of autonomous guided vehicles(AGVs), an observer-based $H_\infty$ preview controller is proposed. Firstly, the path tracking error model of the AGV is converted into a linear variable parameter(LPV) model. Then, a preview model of the path curvature is established, and an augmented system is constructed by combining the path tracking model and the preview model. Considering the influence of sensor measurement noises on the path tracking accuracy, an observer-based $H_\infty$ state-feedback controller is designed, the design conditions of which are converted into solving a set of linear matrix inequalities(LMIs). Finally, the results of simulation utilizing Simulink/CarSim show that, the proposed $H_\infty$ preview controller receives better path tracking results compared with existing methods, where measurement noises are attenuated effectively.