Program

With a growing number of national security challenges, it is imperative that the implementation of Foundation Digital Twins is fully explored and understood in the context of AI. Designing GeoAI systems that will scale with greatest challenges will require engaging the frontlines and visionary societal perspectives for guidance. Join us in this session for a forward framing of the geospatial AI strategy as a bridge toward uncovering unlimited possibilities for impacting national security mission challenges and challenges for our nation’s benefit. 

GEOINT Operational Efficiency will depend on building scientifically grounded and continuously updated digital twins using a dynamic multi-datum geodetic reference framework. The integration of heterogenous global datasets such as GRACE, SWOT, and ICESat-2 into interoperable, geodetically grounded platforms, holds the key to unlock many GEOINT application challenges. In this session we will deep dive into the challenges and visionary solutions to align digital twin platforms with global reference frames and geodetic datums such as ITRF, WGS 84, NATRF2022, and EGM2008. We will discuss why aligning AI systems with global datums and frames, gravity models, and Earth Orientation Parameters (EOPs) is an enabler know the Earth. Panelists will lead the way with grand challenges and visionary solutions toward how these systems can be scaled to support precision, interoperability, and long-term model reliability.

The session will spotlight the next generation of geospatial Foundation models that integrate geospatial knowledge, physical processes, temporal dynamics, and human actions to support pressing energy, security, and environmental challenges. We will take inspiration from the NVIDIA Cosmos – a platform for training general purpose World Foundation Models for Physical AI. Physical AI is characterized as an AI system equipped with sensors and actuators: the sensors enable the system to observe the world, and the actuators allow for interacting with and modifying the world. While GeoAI systems are advancing significantly due to vast amounts of Earth observations and compute scaling in the past two decades, its coupling with physical processes, temporal dynamics, and human actions to account for near real-time and real-world updates is lagging. Join us for an open discussion on general-purpose Cosmos-like World Foundation Models for GeoAI and agents to support Geospatial Foundation Digital Twins. We openly ask, how do we go about developing World Foundation Models that incorporate structured reasoning to understand physics and space-time alignment? What role (if any) will virtual AI replicas play in advancing GeoAI agents?

Toward Enterprise-scale HPC to enable real-time analytics and massive data fusion and analytics. Also considering cloud-native platforms, and emerging quantum architectures to build scalable computing infrastructure for catalyze the development of scalable AI models and workflows.