The National Aeronautics and Space Administration (NASA) partnered with many excellent pioneers of artificial intelligence, including IBM, Google, and Intel, to apply advanced computer algorithms to aid in space science problems.

Machine learning talks about widely used tools and algorithms that help computers to learn from data, categorize objects accurately, and make predictions. Though it allows technology companies to recognize faces in images and predict movies we enjoy, some scientists recognize the possibilities of applying this in space.

NASA’s Frontier Development Lab Teams and Technologies

An astrobiologist from NASA’s Goddard Space Flight Center in Maryland, Giada Arney, together with her colleagues, aspires to utilize machine learning in collecting data for them to use for future observatories and telescopes, like the James Webb Space Telescope from NASA. She believes that machine learning can help her understand the complex data of these essential technologies.

To address this and similar other concerns from other experts, NASA developed a four-year program in partnership with NASA’s Ames Research Center and SETI Institute, which gathers talented people to propel the abundancy of breakthrough technology development.

For eight weeks, the FDL unites pioneers of technology and space each summer to work on and design computer codes. It forms teams of young doctoral students with some professionals from the world’s widely known biggest technology companies and academic communities.

Some of the FDL teams established include:

  • Arney and Shawn Domagal-Goldman

In 2018, they coached an FDL team to develop a machine learning technique for scientists to study atmospheres of planets beyond our solar system, exoplanets, to reduce costly investigation expenses.

Today, they also introduce a neural network technique that can solve highly complex problems similar to how our brains work.

These graduate students from the Universities of Oxford and Central Florida, respectively, led the study that tests the “Bayesian” neural network against the “Random Forest” ML technique. Though they reported its accuracy, it also reports its certainties about its predictions. This feature is recognized as significant by Domagal-Goldman himself.

Succeeding FDL Technologies and Prospects for the Future of Artificial Intelligence

In 2017, FDL participants designed machine learning programs that generated 3D Models of near-approaching asteroids as soon as four days. It paved the way for NASA to locate and block threatening asteroids approaching the Earth.

3D Model of a Near-approaching Asteroid - THESIS.PH

These FDL techniques will be publicly available and show promising applications for sophisticated algorithms to data collected by NASA’s mission. As stated by NASA’s heliophysicist Madhulika Guhathakurta, utilizing these tools fills gaps caused by scarcity in their human resources.

3D Model of a Near-approaching Asteroid - THESIS.PH

Guhathakurta is the key architect in the FDL program. In 2018, she supported the team to resolve NASA’s Solar Dynamic Observatory’s (SDO’s) malfunctioning sensor that functions to report data about EUV radiation levels. Along with doctoral students and their mentors, they generated a virtual sensor that supplied missing data from the broken sensor.

Arney concluded that these AI methods are far from replacing humans due to more verifications needed. Diamond, the SETI head, and Domagal-Goldman visions incorporating these virtual AIs with spacecraft, making real-time science decisions possible.

Do you agree with Arney that artificial intelligence methods cannot replace humans as far as space exploration is concerned? Talk to me in the comments.

LEAVE A REPLY

Please enter your comment!
Please enter your name here