Design of an Intelligent Driver-Passenger Monitoring System for Metro Trains Based on Device-Edge-Cloud Integration

With the ongoing transition of urban rail transit towards intelligent systems, smart crew management has become vital to enhancing operational safety and efficiency. Addressing the issue of insufficient monitoring of driver operations in violation of rules and abnormal passenger conditions in existing crew management systems, this paper proposes a smart driver-passenger monitoring system featuring device-edge-cloud integration for metro trains. This system consists of three layers: a device layer for condition perception, an edge layer for intelligent analysis, and a cloud layer for monitoring management. It employs deep learning-based InsightFace and YOLOv5 object detection models for driver identity authentication, as well as for detecting violations including fatigue driving and distracted driving, and abnormal conditions in passenger compartments. The system assigns differentiated safety constraint factors and penalty mechanisms based on the severity of impact from driver violations on operational safety, to quantify safety risks of detected violations and guide response actions. The system alerts the driver through an onboard voice alarm system to intervene at the initial stage of any detected violation. If an alarm persists, the information is uploaded to the cloud layer, triggering manual intervention by ground crew members. Furthermore, a comprehensive driving operation database covering all trains and drivers is built at the cloud layer of monitoring management, providing data support for smart crew management. Onboard test results demonstrate that the system can effectively monitor various potential hazards related to driving safety in real time, significantly enhancing the standardization of crew management and the safety of train operation.