Frontier Development Lab (FDL) is an applied-AI research program that pairs machine-learning specialists with domain scientists to tackle challenges in space science and Earth science, under the banner “Applied Artificial Intelligence for Space for All Humankind.” It is run as a division of Trillium Technologies in partnership with space agencies and technology companies, and works closely with NASA and the European Space Agency.
FDL is built around short, intensive research sprints in which small mixed teams attack a focused problem over a matter of weeks, supported by cloud computing and GPUs from partners such as Google Cloud and NVIDIA. Its specialized labs cover areas like heliophysics, lunar exploration and Earth observation, and it collaborates with government partners including the US Geological Survey and the Department of Energy.
One of its better-known results is DAGGER, a deep-learning model developed at FDL that predicts geomagnetic storm impacts worldwide about 30 minutes ahead. FDL illustrates a common pattern in modern AI for science: bring outside machine-learning expertise to bear on agency data and instruments through compressed, well-resourced collaborations rather than long-running in-house projects.