Research on Target Detection Algorithms for AGV Under Adverse Weather Conditions

With the rapid development of intelligent manufacturing and smart logistics, object detection has become increasingly important in automated transportation systems, including automated guided vehicles (AGVs), warehouses, production workshops, and distribution operations. However, under adverse weather conditions, existing object detection methods often suffer from degraded performance because object features become blurred or less distinguishable, resulting in reduced detection accuracy. To address this issue, this study proposes an improved object detection algorithm for adverse weather conditions based on YOLOv8n. Specifically, the SimAM attention mechanism is introduced into the backbone network to enhance feature representation. An LCAHead detection head is designed to improve cross-layer feature fusion. In addition, the Wise-IoUv1 loss function is used to replace CIoU, contributing to more stable training and improved convergence. Finally, channel-wise distillation is applied to further enhance detection accuracy without increasing inference cost. Experimental results on the test set show that the proposed method achieves an mAP@0.5 of 50.8%, representing a 7.6% improvement over YOLOv8n, while maintaining an inference speed of 128 FPS.