Research Experiences for Undergraduates (REU), NSF NHERI 2024: Graph Network-Based Simulators and Point-E for 3D Natural Hazard Simulations

This poster, presentation, and paper focus on the integration of two machine learning (ML) methods: Point-E, a text-to-3D point cloud generator, and a Graph Network-based Simulator (GNS) for simulating natural hazard flows. The developed code can be used as a starting point for surrogate modeling of debris flows using the material point method (MPM), Point-E and GNS. Through this project, models are created that can be used as-is or fine-tuned and further improved to enable efficient modeling of hazard-induced flows. Its distinction lies in the integration of a 3D generative model with a surrogate simulator that is relevant to researchers, engineers, and scientists in AI and natural hazards engineering.