Adaptive Integration of BIM and Navisworks for Real-Time Clash Detection Using the XGBoost Algorithm
Odey Alshboul, Ali Shehadeh
Abstract
Integrating building information modeling (BIM) with advanced clash detection technologies, exemplified by Navisworks, has become crucial for effectively managing complex construction projects across architectural, civil, and mechanical, electrical, and plumbing (MEP) disciplines. Traditional clash detection methods remain reactive, leading to costly delays and inefficiencies. This study addresses this gap by presenting an adaptive integration framework that employs the extreme gradient boosting (XGBoost) machine learning algorithm for real-time predictive clash detection. Through direct interfacing with BIM models, our system offers dynamic updates and proactive conflict anticipation, allowing for instant corrective actions. Performance evaluation across diverse BIM projects shows that our framework significantly improved accuracy (0.62 to 0.92), recall (0.65 to 0.93), F1 score (0.65 to 0.92), and receiver operating characteristics (ROC) (0.66 to 0.91). These advancements not only optimize project coordination and efficiency but also dramatically lower the frequency of costly delays and rework, setting a new benchmark in automated clash detection within the construction field. Future research should discover the integration of extended reality (XR) technologies for immersive clash visualization, further enhancing collaboration and decision-making in BIM workflows.