AI-Powered Flash Flood Early Warning System Developed by San Antonio Researchers
Researchers at the University of Texas at San Antonio have developed a self-sufficient warning system powered by solar energy and artificial intelligence to detect street-level floods.
Researchers at the University of Texas at San Antonio have designed a solar-powered, AI-driven flood warning system capable of detecting dangerous street-level water accumulations caused by sudden rainstorms.
Purpose of the Project's Development
Flash floods can emerge within minutes during sudden rainstorms across Texas. In response to this situation, the developed system aims to detect dangerous water accumulation on-site and rapidly.
Technical Specifications of the System
Led by Assistant Professor of Electrical Engineering Chen Pan, the team produced a prototype combining solar energy harvesting, multi-sensor tracking, long-range wireless radios, and on-device machine learning.
Artificial Intelligence and Sensor Technology
The prototype integrates temperature, humidity, light, precipitation, and optical water level sensors. Through the use of TinyML, the device runs machine learning algorithms on small microcontrollers to predict localized flood risk with 98.82% accuracy.
Data Transmission and Cost
When rising water is detected, the system transmits lightweight signals over long distances via LoRa technology without the need for cellular towers. The total cost of a single prototype station ranges between 150 and 220 dollars.