TLDR: Pioneering research at Texas A&M University, in collaboration with Meta, is leveraging advanced AI tools to significantly enhance disaster response and preparedness. These initiatives aim to provide rapid damage assessments, real-time insights, and predictive models to mitigate the impact of natural disasters like hurricanes, floods, wildfires, and tornadoes, ultimately saving lives and resources.
COLLEGE STATION, TX – Artificial intelligence is rapidly transforming the landscape of disaster preparedness and response, with groundbreaking research at Texas A&M University leading the charge. A significant partnership with Meta is amplifying these efforts, focusing on developing sophisticated AI systems to augment situational awareness and resilience capabilities across all stages of a weather hazard.
Dr. Ali Mostafavi, a professor of civil and environmental engineering in the Texas A&M University College of Engineering, highlights Texas as a critical hub for natural disasters, experiencing frequent hurricanes, flash floods, wildfires, and tornadoes. He emphasizes that in such events, even a 30-minute improvement in predicting a flooded road or neighborhood could be instrumental in saving hundreds of lives.
Researchers at Mostafavi’s UrbanResilience.AI Lab, in collaboration with Meta, are developing AI models that leverage large language models, such as Meta’s Llama, to create ‘Disaster Management Companion AI systems.’ These systems are designed to analyze vast, disparate datasets with remarkable speed and computing power, a task that would be overwhelming for human analysts. This includes historical flood data, topographic and hydrologic characteristics, real-time rainfall sensor readings, stream gauges, satellite images, location-based cellphone activities, news articles, and even social media posts from residents.
During the mitigation phase, these predictive models can assess which neighborhoods are most susceptible to flooding before a hurricane makes landfall and even anticipate evacuation patterns. In the active response phase, AI can monitor neighborhood-level evacuations, power outages, and property damages, providing near real-time insights to decision-makers for efficient deployment of search and rescue personnel and other critical resources.
Laura McGorman, Meta’s Director of AI for Good, stated that the company has been working closely with Texas A&M researchers to build AI models that utilize social media data to better predict and respond to natural disasters. She noted that a tool has been developed for first responders in Harris County, Texas, to improve situational awareness using AI based on data from previous Texas disasters.
Meta’s AI, powered by its Llama model, offers several practical features: generating hurricane supply lists and preparation tips tailored to local needs, guiding users through evacuation routes, tracking storms in real-time, and providing hyperlocal insights by leveraging historical data. These tools were previewed for first responders at a workshop in Harris County in May, with the goal of scaling them for use in various disaster scenarios, including wildfires, hurricanes, and earthquakes across the United States.
Beyond the Meta partnership, Texas A&M has also developed CLARKE (Computer vision and Learning for Analysis of Roads and Key Edifices), an AI system that uses drone imagery to rapidly assess damage after hurricanes and floods. Developed by a team led by Tom Manzini and Dr. Robin Murphy, CLARKE can assess a neighborhood of 2,000 homes in just seven minutes, classifying building and road damage and even generating Google Maps-style route planners to help responders avoid impassable roads. This system has already proven its value during the 2024 hurricane season in Florida and Pennsylvania.
Also Read:
- AT&T Unveils Geo Modeler AI System for Enhanced Network Resilience Amidst Extreme Weather
- Cal State LA Secures $1.4 Million NASA Grant for Advanced AI Wildfire Prediction System
By making these AI tools free and open source, Meta and Texas A&M hope to foster broader adoption and continued innovation within the research community and local governments, ensuring that the best available technology is leveraged during crises.
