For the first time, a combination of sky-watching cameras and drones has been used to find a meteorite, opening up a new era in our access to information about the Solar System.
We spend billions of dollars to send spacecraft to asteroids to bring back pieces to analyze, but sometimes those pieces make things easier, coming to us instead in the form of meteorites. Unfortunately, most of the space rocks available to study have spent long enough on the ground they're not in quite the pristine state we would like.
Planetary scientists have set up networks of cameras to track the paths of incoming objects in the hope of finding meteorites more quickly. Last year, they took this to a new level by enlisting drones. Their success in pinpointing the location and finding the meteorite in the Australian outback has been announced in a preprint paper under peer review by the Astrophysical Journal Letters. To find the small rock in a rather large desert required not only path mapping, drone technology, and persistence, but also machine learning to train the drones in what they were looking for.
Just four days after the Desert Fireball Network (DFN) observatory first captured the bright fireball entering the Earth's atmosphere, scientists found the 70-gram meteorite in Western Nullarbor on the Lintos Paddock of Kybo Station in Western Australia, just 50 meters (164 feet) from the calculated path.

Some meteorites are so common their value is fairly low, but rare varieties are truly precious to science. One unusual specimen discovered in Murchison, Victoria, in Australia, in 1969 provided so much insight into the Solar System it had an entire book written about it, as well as sparking numerous papers and a local tourism industry. The value is enhanced when we have a good idea of the rock's path through the atmosphere, enabling us to map its previous orbit and sometimes match it to the asteroid it was once part of.
The Desert Fireball Network was established to both increase the chance of recovering meteorites and to provide multiple images of their flight. It takes advantage of the vast areas of Australia free from artificial light that would disrupt the images, and few plants to hinder the search for sky rocks once they land.
On April 1 last year the DFN recorded a bright meteor that was no April fools' joke. Curtin University graduate student Seamus Anderson and co-authors observed the light was bright enough to make it likely that part of the incoming rock would make it to the ground and used the camera images to map out a search zone 5.1 square kilometers (2.0 square miles) in size on the vast Nullarbor Plain in Western Australia.
“A camera-fitted drone flies over and collects images of the fall zone, which are transferred to our field computer where an algorithm scans each image for meteorites and features that resemble them,” Anderson explained in a statement. “Although our algorithm was ‘trained’ on data collected from past meteorite searches, we brought with us previously recovered meteorites and imaged them on the ground at the fall site, to create local data with which to further train the algorithm.”

The work required about 10 percent of the labor usually involved in a meteorite search – an important saving when searching for rocks in a remote location under the burning Australian Sun. The process was not entirely smooth though, since the paper notes “We have encountered false positives such as tin cans, bottles, snakes, kangaroos, and piles of bones from multiple animals.”
Anderson told IFLScience this was the eighth meteorite recovered by the DFN, but the first using drones. Preliminary analysis suggests it is an H-chondrite, a fairly common meteorite type, but even if this is confirmed the work should clear the way to finding rarer versions.
Anderson told IFLScience the likelihood of anything making it to ground is calculated based on the brightness of the flash and the lowest images returned. “In this case it suggested there was probably something there, but it would likely be on the small side,” he said.