Abstract: To mitigate the increased disturbance induced by parametric uncertainties during turning maneuvers, which severely degrade path-tracking accuracy, and to compensate the non-real-time data updates caused by the low update frequency of GPS-RTK signals in the autonomous navigation system, this study designs a state observer and a sampled-data controller. By establishing a deviation model for the unmanned vehicle and introducing coordinate transformation, its kinematic model is derived. A state observer is constructed to estimate the states of the system, and a sampled-data output feedback controller is designed, with an analysis of the system stability. Finally, simulations and experiments demonstrate that the designed controller can stabilize the system and outperforms the Stanley controller. Under the same parameter conditions, compared to the Stanley controller in the experiment, the designed sampled-data controller significantly improves the navigation accuracy. After the vehicle is launched, the maximum lateral offset for straight path tracking does not exceed 10 cm, and the lateral offset for curved path tracking does not exceed 50 cm, which can well meet the navigation requirements of the driverless vehicles.