Large asteroids whiz past near the earth at regular intervals, for example on March 21 last year. The bumblebee with a diameter of 900 meters was discovered 20 years ago. But they are not always discovered in such good time.

The B612 Foundation was formed in 2002 in response to US authorities not taking the asteroid threat seriously enough. Initially, the goal was to raise money for its own telescope for the purpose, but after Nasa had taken on the task, the private organization chose to redirect its funds.

In 2017, the B612 Foundation launched the Asteroid Institute. There, they work to discover asteroids that are hidden in the data that astronomers collect for other research purposes. Now they have a new tool, the institute writes in one press release.

Then the object is flagged as an asteroid

In collaboration with the University of Washington, an algorithm has been developed that can thin the huge surface and identify potentially dangerous asteroids. One problem was that the basic data lacked so-called tracklets, ie a series of observations of an object made during a night. However, the new software can anticipate this and display a trajectory based on a single point of light.

The software has been named tracket-less heliocentric orbit recovery, abbreviated Thor. It makes assumptions about direction and speed, among other things with the support that asteroids follow certain orbits.

The system creates test tracks that are followed up by new observations, and after five to six such observations within a few weeks, the object is flagged as a potential asteroid.

The Noirlab source catalog (NSC) consists of close to 68 billion observations made by the National Optical Astronomy Observatory’s telescope 2012–2019. As a first demonstration, Thor went through the data from 30 days and then found 104 previously unknown objects. This required heavy cloud computing and cloud storage, provided by Google.