Umeå University's logo

To enhance on-site safety management, detecting near-misses in a timely and accurate manner is crucial. However, traditional monitoring methods that rely on inspectors are time-consuming, labor-intensive, and limited in their ability to detect three-dimensional (3D) incidents. Digital twin technology presents a promising solution by integrating real-time monitoring, simulation and interactivity capabilities in three dimensional virtual environments. Therefore, we propose a real-time digital twin-driven near-miss detection system to facilitate safety management at construction sites. Using advanced stereovision techniques, the system reproduces the construction site as a digital twin model, enabling the real-time 3D monitoring of near-miss events. Additionally, the system incorporates an early warning mechanism to alert workers and managers of potential safety hazards. To demonstrate the effectiveness of the proposed system, a lifting project under various exposure conditions was investigated. The results demonstrate the system's ability to detect and visualize near-misses in three-dimensional space and provide proactive warnings of potential hazards.