<![CDATA[Ensuring safety and efficiency at warehouse intersections has become increasingly vital in the era of automation and intelligent logistics. This study proposes a vision-based anti-collision traffic management system tailored to the dynamic warehouse environment. By combining YOLOv5 object detection with a real-time microcontroller-based actuation system, the system detects and prioritizes movement between forklifts and pedestrians. Four webcams positioned at warehouse intersections transmit visual data to a Raspberry Pi 4, which performs object detection and decision-making based on predefined priority rules. Actuation is executed via Arduino Uno and Nano for signaling "GO" or "STOP" using running text displays and buzzers. The system achieved a mean Average Precision (mAP) of 94.7% and a response latency below 500 milliseconds, enabling safe, real-time operation. Experimental results demonstrated high detection accuracy and effective prioritization logic in four operational scenarios. Compared to traditional sensor-based systems, this approach is more cost-effective, scalable, and adaptable to real-world warehouse conditions. The novelty of this research lies in its integration of modular computer vision, decentralized microcontroller-based actuation, and intelligent traffic prioritization within a low-cost architecture"”features rarely combined in prior industrial safety solutions. Beyond warehouse environments, the proposed system is highly adaptable to other industrial settings such as factories, loading docks, and construction zones, where dynamic human"“machine interactions demand similar real-time visual monitoring and signaling. This work lays a foundation for smart industrial ecosystems, with future extensions toward IoT integration, predictive analytics, and reinforcement learning"“based decision-making.]]>
Copyrights © 2025
