Smart manufacturing systems require efficient and real-time quality control mechanisms to ensure high product reliability and minimize production losses. Traditional cloud-based inspection systems suffer from high latency, bandwidth limitations, and delayed decision-making, while edge-only solutions are constrained by limited computational resources. To address these challenges, this paper proposes an Edge–Cloud Collaborative Framework for Real-Time Quality Control, integrating edge computing for low-latency defect detection and cloud computing for large-scale analytics and model optimization. In the proposed system, real-time data from industrial sensors and vision systems is processed locally at the edge for immediate defect detection using a lightweight deep learning model, while the cloud layer performs periodic model training, optimization, and global decision support. A dynamic task offloading strategy is implemented to balance computational load between edge and cloud based on latency, bandwidth, and resource availability. The framework is evaluated using the MVTec AD dataset on an edge device (Jetson Nano) integrated with a cloud platform. Experimental results demonstrate that the proposed system achieves an accuracy of 97.5%, precision of 97.1%, recall of 96.8%, and F1-score of 96.9%, with an average inference latency of 45 ms, significantly outperforming traditional cloud-only systems (latency ~150 ms) and edge-only systems (accuracy ~94.5%). Additionally, the collaborative approach reduces bandwidth usage by approximately 40% through local preprocessing at the edge. These results confirm that the proposed edge–cloud collaborative framework provides an effective balance between low latency, high accuracy, and efficient resource utilization, making it highly suitable for real-time quality control in Industry 4.0 smart manufacturing environments.
Malik et al. (Thu,) studied this question.