Frontier Development Lab transforms space and earth science for NASA with Google cloud artificial intelligence and machine learning technology
Frontier Development Lab’s goal is to apply synergies between physics, simulation and machine learning to problems important to space science and humanity
Business AnnouncementIMAGE: FRONTIER DEVELOPMENT LAB 2021 CHALLENGES view more
CREDIT: SETI INSTITUTE
August 26, 2021, Mountain View, Calif., – Frontier Development Lab (FDL), in partnership with the SETI Institute, NASA and private sector partners including Google Cloud, are transforming space and Earth science through the application of industry-leading artificial intelligence (AI) and machine learning (ML) tools.
FDL tackles knowledge gaps in space science by pairing ML experts with researchers in physics, astronomy, astrobiology, planetary science, space medicine and Earth science. These researchers have utilized Google Cloud compute resources and expertise since 2018, specifically AI / ML technology, to address research challenges in areas like astronaut health, lunar exploration, exoplanets, heliophysics, climate change and disaster response.
With access to compute resources provided by Google Cloud, FDL has been able to increase the typical ML pipeline by more than 700 times in the last five years, facilitating new discoveries and improved understanding of our planet, solar system and the universe. Throughout this period, Google Cloud’s Office of the CTO (OCTO) has provided ongoing strategic guidance to FDL researchers on how to optimize AI / ML , and how to use compute resources most efficiently.
With Google Cloud’s investment, recent FDL achievements include:
- Allowing lunar scientists to see into the permanently shadowed craters of the lunar poles to help NASA develop traverse plans for robotic and human exploration.
- Helping clinicians understand the causal factors in a target cancer in a way that respects patient privacy.
- Enabling researchers to search NASA's petascale archives of data without human labels.
- Helping astrophysicists survey the starspots on distant stars to determine their suitability for life.
- Updating U.S. hydrology maps everyday to help the U.S. Geological Survey understand the U.S. water supply to develop better predictive models for droughts and floods.
"Unfettered on-demand access to massive super-compute resources has transformed the FDL program, enabling researchers to address highly complex challenges across a wide range of science domains, advancing new knowledge, new discoveries and improved understandings in previously unimaginable timeframes,” said Bill Diamond, president and CEO, SETI Institute. “This program, and the extraordinary results it achieves, would not be possible without the resources generously provided by Google Cloud.”
“When I first met Bill Diamond and James Parr in 2017, they asked me a simple question: ‘What could happen if we marry the best of Silicon Valley and the minds of NASA?’” said Scott Penberthy, director of Applied AI at Google Cloud. “That was an irresistible challenge. We at Google Cloud simply shared some of our AI tricks and tools, one engineer to another, and they ran with it. I’m delighted to see what we’ve been able to accomplish together - and I am inspired for what we can achieve in the future. The possibilities are endless.”
FDL leverages AI technologies to push the frontiers of science research and develop new tools to help solve some of humanity's biggest challenges. FDL teams are comprised of doctoral and post-doctoral researchers who use AI / ML to tackle ground-breaking challenges. Cloud-based super-computer resources mean that FDL teams achieve results in eight-week research sprints that would not be possible in even year-long programs with conventional compute capabilities.
“High-performance computing is normally constrained due to the large amount of time, limited availability and cost of running AI experiments,” said James Parr, director of FDL. “You’re always in a queue. Having a common platform to integrate unstructured data and train neural networks in the cloud allows our FDL researchers from different backgrounds to work together on hugely complex problems with enormous data requirements - no matter where they are located.”
Better integrating science and ML is the founding rationale and future north star of FDL’s partnership with Google Cloud. ML is particularly powerful for space science when paired with a physical understanding of a problem space. The gap between what we know so far and what we collect as data is an exciting frontier for discovery and something AI / ML and cloud technology is poised to transform.
###
You can learn more about FDL’s 2021 program here.
The FDL 2021 showcase presentations can be watched as follows:
- FDL USA in full - https://youtu.be/viAm8gAlb3g
- Astronaut Health - https://youtu.be/lhe7gPjo7HQ
- Solar Wind - https://youtu.be/5gwMC4Rw5zA
- Solar Drag - https://youtu.be/gEL89pH4PWU
- Automated reporting (NLP) - https://youtu.be/ieA1isaXMnk
- Upscaling Lunar resources - https://youtu.be/op740EPXQ30
- GLM Lightning - https://youtu.be/CnUx2eOz6zc
- Digital Twin Earth - Coasts - https://youtu.be/7wjtcJxRjKk
In addition to Google Cloud, FDL is supported by partners including Lockheed Martin, Intel, Luxembourg Space Agency, MIT Portugal, Lawrence Berkeley National Lab, USGS, Microsoft, NVIDIA, Mayo Clinic, Planet and IBM.
About the SETI Institute
Founded in 1984, the SETI Institute is a non-profit, multidisciplinary research and education organization whose mission is to lead humanity's quest to understand the origins and prevalence of life and intelligence in the universe and share that knowledge with the world. Our research encompasses the physical and biological sciences and leverages expertise in data analytics, machine learning and advanced signal detection technologies. The SETI Institute is a distinguished research partner for industry, academia and government agencies, including NASA and NSF.
Contact Information:
Rebecca McDonald
Director of Communications
SETI Institute
rmcdonald@SETI.org
METHOD OF RESEARCH
Computational simulation/modeling
No comments:
Post a Comment