Frontier Development Lab (FDL) is looking for researchers at the PHD and Post Doc level with expertise in machine learning/artificial intelligence/data science as well as those with expertise in science domains including heliophysics, planetary science, Earth science, disaster management, astrophysics and astronaut health. The deadline to apply is April 6, 2020.
FDL is an applied artificial intelligence research initiative that pairs researchers from the science domains with data scientists for an intense 8-week sprint to apply AI/ML to challenges important to NASA’s science and exploration goals – and all humankind.
FDL runs between June 22nd and August 14th, 2020, hosted by the SETI Institute and NASA Ames in Mountain View, CA.
NASA provides funding for FDL through a cooperative agreement with the SETI Institute. Compute time, challenge-specific data and advisory support is provided by NASA, Google, Intel, IBM, Intel, Nvidia, Lockheed Martin, HPE, Kx, MIT, The Mayo Clinic and USGS.
Researchers receive a stipend and are provided with accommodation.
Please note that the challenge list is still provisional and while most challenges are confirmed to go ahead, some may be adapted or moved to 2021 based on capacity. All challenge ideas remain in the system for consideration.
2020 Proposed Challenges:
Heliophysics
1. Tracking the geoeffectiveness of solar storms
Using continuous data gathered by space and ground-based remote and in-situ sensing instruments to predict the impact of individual solar storms on terrestrial systems.
2. Earlier detection of solar wind characteristics
Can we use ML to predict the solar wind (Bz) at any point in the solar system?
Planetary Science
3. Moon Engine: Moon for Good, Phase II
Leveraging the work of the 2019 FDL Moon for Good team to extend their work toward identifying lunar resources of interest in strategic regions as well as producing 3-D high-resolution maps of the North and South Poles to support future lunar exploration missions.
Earth Science
4. The Earth Intelligence Engine: Drought images from the Future
Can DL enhanced simulation acceleration methods and / or generative adversarial networks (GANs) produce predictions and enhanced uncertainty estimations of drought behavior under future climate conditions to inform long-term resilience planning strategies -AKA ‘satellite images from the future’
Disaster Management
5. Lightning and Extreme Weather
Improving the capability to predict extreme weather events arising from convective activity.
6. Floods: mapping inundation extent during a flooding event
Can we support disaster efforts by being able to rapidly map current water extent and differencing against known or typical water, to rapidly identify those regions where flooding is happening, using high temporal satellite imagery and SAR?
Astrophysics
7. Star Check: seeking out unusual stars and planetary systems
Recent NASA space telescope missions have uncovered stars exhibiting bizarre brightness changes over time, meanwhile each new solar system discovered challenges our understanding of how solar systems evolve, but the search space is enormous. Can ML help?
8. Starspots: Stellar surface features from exoplanetary transits
Use exoplanetary transits to obtain information about active regions and starspots on host stars and provide insights on habitability.
Astronaut Health
9. Long duration missions and cancer: a testbed for building causal inference methods
Can we use causal inference methods to understand the molecular basis of cancer in high radiation environments, such as a long duration stay on the Moon or Mars?
FDL’s response to COVID 19
We are watching the situation and following World Health Organization advice closely. We are putting in place contingency processes and plans in the event of continuing limits on local and international travel. We expect FDL to go ahead this summer but will be tailoring our processes, location and working methods (including virtual participation) as needed to safeguard everyone joining us.
