CAREER: OpenGPRxAI: Open-Source, AI-Powered Ground Penetrating Radar Technologies for Real-Time Subsurface Vision

About This Grant

This project aims to significantly advance subsurface imaging by developing advanced, open-source, AI-powered ground-penetrating radar (GPR) technologies. GPR is a non-invasive sensing tool that uses electromagnetic waves to detect and visualize objects beneath the surface, making it invaluable for applications such as precision agriculture, defense, archaeology, civil engineering, and planetary exploration. Despite its broad utility, current GPR technologies face significant limitations, including the lack of standardized datasets and the inability of existing AI models to generalize across diverse systems and environments. This project addresses these challenges by creating innovative solutions that will enable real-time, high-resolution subsurface imaging. The outcome of this project is expected to transform how humans and autonomous systems perceive and interact with the subsurface world. The societal benefits of this work are profound: it will support sustainable water management, improve precision agriculture for global food security, enhance humanitarian demining efforts to save lives, and enable safer infrastructure monitoring. Additionally, the project includes a robust education and outreach component, engaging K-12, undergraduate, and graduate students in electromagnetics and AI through hands-on learning experiences. By fostering interdisciplinary collaboration and launching an open-source online platform, the project will create a global hub for sharing GPR datasets, algorithms, and educational resources, driving innovation and expanding access to cutting-edge subsurface sensing technologies. The research of this project will advance GPR technology through four integrated objectives. First, it will develop far-field and near-field GPR domain transfer frameworks to standardize data collected across different systems, addressing the critical bottleneck of data scarcity and incompatibility. Second, it will create a modular, physics-informed deep-learning AI framework to process GPR data for real-time, high-resolution subsurface permittivity mapping. This AI framework will include models for clutter removal, subsurface medium permittivity estimation, and 2D and 3D imaging, enabling rapid adaptation to diverse GPR systems and environments. Third, the project will validate these AI frameworks in real-world applications, including soil moisture mapping, buried explosive ordnance detection, and underground crop imaging, demonstrating their effectiveness and practical impact. Finally, the project will launch the OpenGPRxAI initiative, an open-source platform for sharing GPR datasets, pre-trained models, system designs, and educational resources. The research employs cutting-edge methods such as deep-learning AI, domain transfer techniques, and simulation-based dataset generation to overcome existing limitations in GPR technology. By addressing challenges in data standardization and model generalizability, the project will significantly enhance the speed, accuracy, and accessibility of subsurface imaging. These advancements will not only benefit diverse applications but also foster interdisciplinary collaboration and innovation, establishing a sustainable research and education community dedicated to intelligent subsurface imaging. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.