NASA’s Asteroid Data Hunter – Machine Learning Algorithm

Spotting asteroids in the night sky is a critical and difficult endeavor. See how crowdsourcing delivered a machine learning algorithm and application that opens this task of asteroid hunting to the citizens of Earth.

Project Details

Spotting asteroids in the night sky is a critical and difficult endeavor. Most asteroids stay in the asteroid belt between Mars and Jupiter. However, some come close to or even impact Earth. Asteroids pose both a possible threat and an opportunity for Earth: they could impact us, causing damage, OR possibly be mined for resources that could help extend our ability to explore the universe.


NASA is the U.S. government agency responsible for civilian space programs as well as aeronautics and aerospace research. NASA, Harvard University in association with Harvard’s Institute of Quantitative Social Science, and Topcoder have established the NASA Tournament Lab (NTL), enabling the Topcoder Community to compete to create the most innovative, efficient and optimized solutions for specific real-world challenges being faced by NASA researchers. Planetary Resources, a private corporation, aims to mine asteroids in order to “bring the natural resources of space within humanity’s economic sphere of influence.” Planetary Resources focuses on extracting from asteroids both (i) water to help create rocket fuel and for consumption is space habitats, and (ii) rare metals such as platinum, which asteroids may contain in massive quantities.


Protecting the Earth from the threat of Asteroid impacts means first knowing where they are. NASA and Planetary Resources — with the help of the Crowd Innovation Lab at Harvard University — wanted to develop an algorithm that would allow the discovery of new asteroids by analyzing images through an image recognition challenge. The goal was to crowdsource an algorithm that is capable of using imagery data from modern telescopes to find more asteroids than had previously been possible


After running 10 contests and awarding $71,000 in prizes, the new algorithm is approximately 15% more accurate than the current method of identifying asteroids in the main belt between Mars and Jupiter. The algorithm is capable of running on a common laptop/desktop in the form of a simple app.

0 %
More accurate than current method

Project Metrics

$ 0