In agricultural operations such as planting, spraying, or harvesting, maintaining precise paths is paramount to prevent crop damage. This paper presents an enhanced GPS-based control architecture customized for a differential-drive agricultural unmanned ground vehicle (UGV). By combining classic control approach with an orientation-first strategy, the proposed architecture offers improved accuracy for the robot navigation. The fusion of these two approaches enables the UGV to navigate through narrow pathways with crops on both sides while minimizing the risk of crop damage. Simulation results using the Robot Operating System (ROS) and gazebo simulator demonstrate the superiority of the proposed control approach over standard controller, ensuring both crop safety and accurate execution of agricultural tasks.