Schedule Control with Automated Progress Monitoring in Modular Construction Sites: Controlled Experiment

Abstract

This study presents a comprehensive controlled experiment designed to evaluate the efficacy of predicting schedule control from automated progress monitoring in modular construction sites. Modular construction requires highly synchronized logistics and assembly processes, making traditional manual progress tracking methods insufficient due to latency and human error. To address this, we implement an automated progress monitoring system integrating computer vision and Internet of Things sensors within a simulated modular construction environment. Through controlled experimental runs, we vary operational conditions such as occlusion levels, lighting, and assembly complexity to test the robust performance of the monitoring system. The gathered real-time spatial-temporal data is utilized to train and validate machine learning models for forecasting schedule deviations. The results demonstrate that the proposed automated system significantly reduces tracking latency and improves predictive accuracy, achieving high correlation with actual ground-truth milestones. Furthermore, the schedule control model successfully anticipates delays up to four cycles in advance, providing project managers with a reliable decision-support mechanism to implement corrective actions. This research offers valuable insights into the transition from reactive progress tracking to proactive schedule control, ultimately contributing to enhanced productivity and minimized schedule overruns in modular construction projects.