As he watched the spacecraft roar towards the stars, he knew the real work wouldn’t happen for another few years.
September 8th, 2016 was a monumental day for UNCOMN’s Ryan Owens (and for NASA). The spacecraft Origins, Spectral Interpretation, Resource Identification, Security, Regolith Explorer (OSIRIS-REx) began its seven-year journey towards the asteroid Bennu and the world began patiently waiting for its return. The main headline for OSIRIS-REx is that the mission will bring back about a Snickers bar worth of asteroid rocks and dust but for Owens, it’s pictures that bring the excitement.
Now he is a Systems Engineer for UNCOMN but while working for the nonprofit CosmoQuest, Owens helped pioneer software that allows researchers to leverage machine learning to more effectively review images taken from space. Using Tensorflow via Amazon Web Services, he and his team created a neural net that will be able to review images of celestial objects, catalog them, and tag them for researchers’ review, allowing researchers to spend more of their time on analysis and extrapolation and not simple data management and cataloging. The work earned him and his team enough attention from NASA to be invited to the actual OSIRIS-REx launch.
OSIRIS-REx arrived at Bennu on December 3rd and began its initial approach. The craft will have to take thousands of images of the asteroid in preparation for its retrieval mission and, in order to get the necessary data, OSIRIS-REx will break an orbiting record today (31 December). At just 1640ft across, Bennu will become the smallest celestial object ever orbited by a manmade craft. Since the asteroid is so small it has almost no gravitational pull, forcing the craft to orbit at a very short distance (.87 miles). Landing on the asteroid isn’t feasible so it will have to make a pogo stick like jump off of Bennu’s surface to grab some rocks and dust to bring back home
Once OSIRIS-REx returns to Earth, those thousands of images will need to be reviewed, cataloged, and documented. That is a monumental task for humans to complete and, until very recently, they had to do the work manually. Now, with the help of machine learning, that has started to change.
Space exploration is not just reserved for astronauts, NASA, and billionaire entrepreneurs. The current work NASA accomplishes is strongly supported by an army of volunteer citizen scientists. They come from all walks of life and devote their time to assist professional scientists in their understanding of the universe. Owens is one of those citizen scientists.
Once the photos of Bennu are made available to the public, Owens and other citizen scientists will be able to run the images through data collection applications online and provide the machine learning neural net with the information it needs to pinpoint the exact size, shape, and distances of thousands of objects on the asteroid. The work won’t be done overnight. But if it could be, where’s the fun in that?