SIGMA Lab announces four core research directions linking below-ground geotechnical data with above-ground infrastructure and project management data on a single GIS and BIM platform.
For the 2026-2029 period, SIGMA Lab focuses on four research directions.
The first applies machine and deep learning to urban geotechnical hazard and risk prediction, using Random Forest, XGBoost, SVM, CNN, LSTM and Transformer models on geotechnical investigation data, monitoring records and InSAR remote sensing time series integrated on a GIS platform.
The second applies deep learning to urban infrastructure asset management, using object detection and semantic segmentation (YOLO, U-Net) to automatically identify pavement cracking and structural deterioration from field and UAV imagery, combined with degradation forecasting on WebGIS and BIM-GIS platforms.
The third applies machine learning to construction project and investment risk management, predicting schedule and cost overrun risk while using natural language processing on project documents and multi-criteria decision making.
The fourth builds the spatial data platform, MLOps pipelines and decision support layer, including shared training datasets, model training and deployment workflows, and a decision support system built on explainable AI.