<p>To address the challenges of reduced navigation accuracy and insufficient steering stability in automated guided vehicles (AGVs) caused by complex obstacle layouts and dynamic environmental changes, this paper proposes a path planning-oriented data fusion differential steering visual neuron PID control method for AGVs. By constructing a multi-sensor system integrating vision, proximity sensors, and a gyroscope, comprehensive environmental and state data of the AGV are collected, providing a multi-source information foundation for subsequent control. A visual neuron dynamic model is employed to fuse and transform multi-source data, generating bounded smooth fusion signals to enhance the accuracy and robustness of state perception. A global path planning model is established with the objectives of minimizing path length and maximizing smoothness, and solved using a particle swarm optimization algorithm, ensuring that the resulting path satisfies motion constraints while balancing efficiency and stability. Finally, a PID controller, driven by the visual neuron output and path planning results, generates differential steering control variables to achieve real-time and precise AGV steering control. Experimental results demonstrate that the proposed method effectively performs global path planning for AGVs and controls trajectory deviation within ± 15 mm. The average control success rate reaches 96.35%, with an average response time of 0.68 s, significantly improving navigation performance and control reliability in complex scenarios. The method exhibits favorable practical application outcomes.</p>