Data integration using deep learning and real-time locating system (RTLS) for automated construction progress monitoring and reporting

Abstract

The shift towards automated progress monitoring using new technologies for efficient delivery of construction projects has received significant attention. The application of vision-based techniques for object recognition and real-time locating system (RTLS) for object localization has been widely studied. However, a single technology cannot provide complete information needed to determine the status of tracked elements on a job site. This paper presents an integrated method for progress monitoring through recognition and localization of elements in construction sites. This method integrates data derived from a deep learning model and Ultra-wideband (UWB) system, and reports each element's ID, location, visual data and capture time. Such information is essential for project managers to assess progress on site. The method is validated in a mechanical room, a challenging environment for RTLS and object recognition models due to signal interferences and occlusions. The findings suggest further research on improving integrated methods for efficient progress reporting.