AI-Based Urban Building Risk and Energy Efficiency Assessment System | NYU Tandon School of Engineering

AI-Based Urban Building Risk and Energy Efficiency Assessment System

Center for Urban Science + Progress
Transportation & Infrastructure,
Urban
Project Sponsor:

Dipendu Bhunia, Professor and Head of the Department, BITS Pilani

 

MENTOR:

Mukund Lahoti, Associate Professor, Department of Civil Engineering, BITS Pilani
Rishabh Chauhan, Industry Assistant Professor at the Center for Urban Science + Progress at NYU Tandon

Authors

Shihan Zhou, Ziqi Liu

Research Question

How can AI/ML and open geospatial data be used to identify building risk (structural or energy-related) and guide retrofitting decisions in urban areas?

Background

Many urban buildings, especially in growing cities, are aging or poorly documented. This project aims to build an AI-powered decision-support tool to evaluate the vulnerability and inefficiency of buildings in urban zones using public datasets. Using satellite imagery, building footprint data (like OSM), and urban heat maps, the model predicts:

  • Structural stress indicators (based on morphology, density, proximity to faults, or vegetation)
  • Energy inefficiency zones (heat loss, poor ventilation, materials)

The project uses open tools like Google Earth Engine and machine learning libraries (TensorFlow, PyTorch) to analyze image and geospatial data. The model outcomes are visualized via a dashboard that city planners or architects can use to prioritize buildings for retrofitting or energy optimization.

This project empowers city agencies and policymakers to build smarter, more sustainable cities using data-driven methods, directly supporting CUSP's mission in urban science and resilience.

Methodology

Techniques and tools used for this project include:

  • Image segmentation using CNNs
  • Heat signature classification
  • GIS-based data visualization (QGIS, Leaflet, Folium)
  • Multi-criteria analysis (e.g., AHP for vulnerability ranking)

The project steps include:

  1. Data gathering and preprocessing

  2. Feature engineering (age, density, location risks)

  3. ML model training and validation

  4. Dashboard/Map-based output visualization

  5. Policy-level interpretation of results

Deliverables
  • Predictive model for building inefficiency/risk
  • Interactive GIS dashboard for urban planners
  • Technical paper or report on AI-ML model performance
  • GitHub repository and dashboard
Data Sources
  • OpenStreetMap (OSM): Building footprints, elevation data
  • Sentinel-2 / Landsat Satellite Imagery: Heat mapping, vegetation index
  • NASA SEDAC Urban Expansion / Land Use Data
  • WorldClim / ERA5 Weather Data: Temperature/humidity
  • UN or World Bank city-scale infrastructure indices