Every day, our planet is bombarded with hundreds of pieces of natural space debris like chunks of comets or asteroids. Some of these rocks survive their journeys through the atmosphere and end up in remote areas or in the oceans, making the task of locating them extremely hard. But now, there's a new tech that scientists can make use of that could be used to hunt them down.
Collecting freshly fallen meteorites, especially those who are monitored, is crucial for understanding where they came from, what's their story and how meteorite impacts could affect our future here on Earth. Even though the researchers' ability to locate meteorite falls is improving, recovering them is still a difficult task due to the large regions obscured by topography and vegetation.
To address that issue, a team of researchers from the University of California has published a new study in which they propose using drones and machine learning to perform automated searches for these alien rocks. So how would it work?
Well, they say that a quadcopter drone will initially conduct a grid scan, taking photos of the strewn field (the region in which meteorites from a single fall are scattered) from a low altitude in order to identify the meteorites. Next, using a machine learning classifier, the images will be analyzed to choose the suitable samples for examination.
In order to test the theory, the team used an off-the-shelf 3DR quadcopter drone and a GoPro Hero4 on a suspected strewn field in Walker Lake, Nevada. They programmed the drone to fly and photograph the region at a constant altitude of 2-6 meters (6.5-20 ft). Even if the software used returned several false positives for previously unidentified rocks, it was able to correctly identify meteorites on a dry lake bed.
This successful test demonstrates that drones are capable of performing autonomous searches for meteorites in large, difficult-to-reach locations. Their capabilities will not only be used to cut down the time spent on searching, but they will also help scientists detect small fragments from fresh meteorite falls. The researchers' whole methodology and findings are documented in their study, which was published in Meteoritics and Planetary Science.
To address that issue, a team of researchers from the University of California has published a new study in which they propose using drones and machine learning to perform automated searches for these alien rocks. So how would it work?
Well, they say that a quadcopter drone will initially conduct a grid scan, taking photos of the strewn field (the region in which meteorites from a single fall are scattered) from a low altitude in order to identify the meteorites. Next, using a machine learning classifier, the images will be analyzed to choose the suitable samples for examination.
In order to test the theory, the team used an off-the-shelf 3DR quadcopter drone and a GoPro Hero4 on a suspected strewn field in Walker Lake, Nevada. They programmed the drone to fly and photograph the region at a constant altitude of 2-6 meters (6.5-20 ft). Even if the software used returned several false positives for previously unidentified rocks, it was able to correctly identify meteorites on a dry lake bed.
This successful test demonstrates that drones are capable of performing autonomous searches for meteorites in large, difficult-to-reach locations. Their capabilities will not only be used to cut down the time spent on searching, but they will also help scientists detect small fragments from fresh meteorite falls. The researchers' whole methodology and findings are documented in their study, which was published in Meteoritics and Planetary Science.