Embedded Files
Data-driven Risk and Resilience Assessment to support Decision Making
Data-driven Risk and Resilience Assessment to support Decision Making
The core questions that guide our research lab are:
How can AI help optimize sequential infrastructure investments under uncertainty and resource constraints?
Can generative and agent-based models capture the complex interplay of hazards and human behavior in resilience planning?
What role do foundational models play in fusing multi-modal, asynchronous data (LiDAR, imagery, sensor data) for real-time risk assessment?
Wildfire risk assessment for road network improvement
Wildfire risk assessment for road network improvement
Wildfires threaten road networks, complicating emergency responses.
Proposed framework integrates wildfire simulation, risk assessment, and retrofitting.
GAN-based model evaluates risks using synthetic and historical data.
Network Performance Tensor combines key metrics for optimal retrofitting decisions.
Tested in Los Angeles, the framework enhances resilience and scalability.
Regional landslide risk assessment for road network improvement
Regional landslide risk assessment for road network improvement
Recent Californian route closures underscore earthquake landslide risks.
Proposes a framework for managing landslide risk to road networks.
Introduces Siamese Graph Convolutional Network (GCN) infused Genetic Algorithm (GA) for efficient optimization of retrofits.
Retrofit policy aims for equitable outcomes across income groups.
Framework applied to Los Angeles, enhances road network resilience.
Paper - link
Edge-based GNN for critical road ranking in a road network
Edge-based GNN for critical road ranking in a road network
Proposed method identifies critical road segments for resilience.
Utilizes Graph Neural Network to assess road importance.
Overcomes existing methods' limitations in compromised networks.
Reduces computational load, suitable for large networks.
Proven effectiveness with synthetic and real-world examples.
Paper - link
Fusing hyperlocal data in capacity expansion planning
Fusing hyperlocal data in capacity expansion planning
Rapid urbanization strains roads, and increases disaster vulnerability.
A capital investment algorithm is proposed for optimal road width expansion.
Novel framework combines mobile mapping, cameras, and lidar.
Utilizes deep learning, and point cloud processing for precision.
Outperforms traditional methods in budget-limited situations.
Enhances hillside road planning and automated feature detection.
Paper - link
Page updated
Google Sites
Report abuse