Chedy Raïssi: an astronomy enthusiast in a NASA program
After six weeks spent at the Frontier Development Lab (FDL), hosted by NASA and the SETI Institute, the adventure is continuing for Chedy Raïssi, a researcher on the Orpailleur project-team. Here is a look back at this programme and its repercussions.
Fond of astronomy since childhood, he was encouraged by an astrophysicist colleague to apply to the NASA Frontier Development Lab programme during a conference and exchange on the automatic detection of stellar objects.
A multidisciplinary project
A first for NASA on this subject, the high-security FDL project brought together 12 planetary and machine learning scientists over a period of six weeks to combine their talents and find correlations between their respective fields in order to use technological advances to improve ways to defend the Earth’s lines of defense from asteroids.
The project sought to provide an environment particularly conducive to scientific cooperation, not only between the teams behind the project, but with universities and local companies as well, and enable close interaction with expert mentors.
Three teams tackling three different subjects
The 12 finalist scientists were divided into three teams, each assigned to tackle a specific problem. The first team investigated how drones equipped with cameras could be used to locate and identify meteorites that have crashed on Earth. The second team used machine learning algorithms to simulate near-Earth objects and systematize their results to provide the best solutions to specific situations. Despite the entertainment value of the disaster movie "Armageddon,” J blowing up an asteroid is not particularly recommended, as doing so would create thousands of pieces of rogue debris that could fall to Earth.
Chedy’s team worked on characterizing the shapes of asteroids using radar observations modeled in 3D. In order to be able to land on an asteroid, divert its course, or — in a more distant future — drill into it, scientists need to know what it looks like and what its composition and axis of rotation. Currently, it takes between four and six months of nonstop human intervention to obtain a viable asteroid model from a radar observation. Chedy’s team sought to improve these methods via deep learning networks. Its results — a decrease to only one or two weeks, more reliable findings, and less human intervention — are more than promising.
The FDL project may be over, but cooperation goes on
Despite its limited time frame, the FDL project provided the teams and institutes with enough convincing results to allow them to carry on their cooperation and further explore their findings and research. But the challenge now is how to do so despite each person’s geographic differences and individual projects. Nevertheless, the project continues to live on through articles published in scientific journals, meetings, and discussions. Needless to say, we haven’t heard the last of this fascinating project!
To learn more about the results of Chedy’s Frontier Development Lab team, see https://www.youtube.com/watch?v=-Mz4q1XP6ug
For the findings of the other teams, visit http://www.frontierdevelopmentlab.org/#/results