Key Takeaways:
- AI helps discover 27,000 new asteroids previously missed in telescope images.
- The new asteroids are mostly located in the asteroid belt between Mars and Jupiter.
- AI tool called THOR can analyze up to 1.7 billion light dots in a single telescope image.
- AI is being used by more astronomy teams to discover new asteroids.
- The upcoming Vera C. Rubin Observatory is expected to find up to 2.4 million asteroids in its first six months of operation.
We now have a catalog of the more than 27,000 asteroids in our solar system that were missed in earlier telescope images, all because of a new AI-powered algorithm. The instrument, according to the scientists who made the discovery, makes it simpler to locate and track millions of asteroids, including those that could be dangerous and one day strike Earth. It is for those threatening space rocks that the world would need years of advance warning before trying to deflect them away from our planet.
Over the past 200 years, scientists have cataloged over 1.3 million rocky shards in the asteroid belt between Mars and Jupiter, where the majority of the newly discovered asteroids are located. The most recent find, which took place over the course of five weeks, also includes roughly 150 space rocks whose paths take them through Earth’s orbit; to be clear, though, none of these “near-Earth asteroids” appear to be headed straight for Earth. Others are Trojans that trace the path of Jupiter around the sun. The official organization in charge of asteroid discoveries, the Minor Planet Center of the International Astronomical Union, has not yet received or approved observations of these asteroids.
Conventionally, astronomers look for new asteroids by repeatedly examining certain areas of our sky using telescope images that are taken several times a night, typically every few hours. The background stars, planets, and galaxies are constant from image to image, but asteroids are identified as observable moving spots of light that are flagged and confirmed. From there, these asteroids’ orbits are calculated and tracked.
During a discussion on the discovery early last month, Ed Lu, executive director of the Asteroid Institute and co-founder of the B612 Foundation, said, “This is really a job for AI.” According to Lu, artificial intelligence (AI) tools created for asteroid searches are already getting close to human capabilities: “I think we’re gonna quickly surpass that over the next few weeks.”
The Tracklet-less Heliocentric Orbit Recovery, or THOR, algorithm that Lu’s group created examined more than 400,000 sky archival photos that were kept up to date by the National Optical-Infrared Astronomy Research Laboratory, or NOIRLab. The algorithm can begin operating as soon as there are roughly five observations within a 30-day period linked to the same region of the sky. It can analyze up to 1.7 billion light dots in a single telescope image because it was trained on a large dataset. Its purpose is to locate and link a light point in one sky image to another in order to determine whether or not both specks depict the same object. Usually, this means that an asteroid is passing through the space, according to a description of the algorithm by the B612 Foundation.
Lu said, “We don’t own a telescope, we don’t operate a telescope,” during the conversation. “We’re doing this from a data science perspective.”
According to a statement released by the B612 Foundation on Tuesday, April 30, the scientists scaled their algorithm using Google Cloud, whose computational power and data storage services made it easier for the scientists to test out thousands of orbits of asteroid candidates.
“Not only can we find asteroids in datasets that were never meant for it, but we can make every other telescope in the world better at finding asteroids,” Lu said in his presentation. “It’s a change in how astronomy is done.”
Using THOR, the same scientific team found 100 asteroids in 2022 that were missed in previous telescope images. AI has also been used by other astronomy teams to discover new asteroids. For example, just two weeks ago, the Hubble Space Telescope’s archived images revealed 1,000 new asteroids, thanks to the efforts of citizen scientists who led the algorithm’s training. A 600-foot-wide (180-meter-wide) space rock was discovered in July of last year by HelioLinc3D, software used to search for near-Earth asteroids. The rock is predicted to approach Earth at a distance of 140,000 miles (225,000 kilometers). That is closer than the moon’s average distance from our planet.
Over 2,000 of these “potentially hazardous asteroids” have been identified so far, and scientists believe there may be another 2,000 out there. One of the goals of the upcoming Vera C. Rubin Observatory in Chile, for which the asteroid-hunting HelioLinc3D software was created, is to detect these space rocks in an effort to support planetary defense.
The 8.4-meter telescope, which is scheduled to start operations next year, will take images of the southern sky every night for at least a decade, each image covering 40-full-moons of area. This cadence, according to scientists, could aid the observatory in finding up to 2.4 million asteroids in its first six months of operation—double the number currently cataloged—thanks to AI-based software like THOR and HelioLinc3D.
