It’s a cold winter night in the Western Australian outback. The wind is blowing, the cicadas are buzzing, and far above the remote landscape, a meteorite soars through the sky in a fleeting blaze of glory.
The fireball sparks a wave of excitement amongst researchers at Curtin University’s Desert Fireball Network (DFN), who could use the meteorite to investigate mysteries surrounding the formation of our solar system.
Did the meteorite survive its fiery entrance through Earth’s atmosphere? Is it large enough to find? And if so… where exactly did it land?
The meteorite – a black rock, no bigger than the palm of a hand – could have landed anywhere within the hundred-hectare search area, about 400 km northeast of Kalgoorlie. Trying to find it is like trying to find a needle in a haystack.
So, equipped with a drone, the power of the ARDC Nectar Research Cloud, a lot of patience, and a trusty field truck affectionately named Rhonda, a small team of researchers from the DFN set out into WA’s Central Desert to do the impossible: find it.

Drones: A Meteorite Hunter’s Companion
The meteorite-hunting process begins with DFN’s network of digital observatories – fully autonomous cameras positioned across Australia that continuously monitor over 3 million square kilometres of the night sky. When a meteorite passes overhead, researchers use images from the cameras to estimate where it may have landed based on its trajectory and other factors, like the wind.
Until recently, the next step would involve a group of 6–10 people, spaced about 5 metres apart, manually probing the area like a police line search. This method is highly labour-intensive, often taking weeks or even months of human effort to find the specimen, if it was ever found at all.
Dr Hadrien Devillepoix, lead scientist at the DFN at Curtin University, knew that there had to be a more efficient way.
“A couple of years ago, we had a PhD student, Seamus Anderson, who’s now doing a postdoc at NASA. We basically gave him that task: to find meteorites with drones,” he said. “At the end of 2021, he made it work – using drones to do the automated flying, then processing the images through machine learning algorithms.”
Meteorites vs. Meteowrongs
Using the drones allowed the researchers to thoroughly survey the area by capturing high-resolution images, which significantly reduced human search time.
However, that left the researchers with a different problem: they’d often end up with tens of thousands of images of potential meteorite candidates, since machine learning algorithms can’t always tell real meteorites apart from other small, black objects, like spider holes, gumnuts, or kangaroo poos, jokingly dubbed ‘meteowrongs.’

Piecing Together the Clues with Nectar
To make the process more efficient, the DFN developed the Drone Meteorite Search Platform, a web application that processes all the photos collected by the drones, identifies which items might be meteorites, and enables humans to select only the most promising candidates for the field search.
“One big limitation for us in field work is trying to make the most of the time. We’re trying to make things as efficient as possible, to offload the work of moving and processing data off-site. So that when they’re in the field, researchers can concentrate on the things that only they can do, like fly the drone and follow up the candidates,” Hadrien said.
The Drone Meteorite Search Platform is powered by ARDC’s Nectar Research Cloud.
“The web application runs on Nectar. We’ve got one virtual machine that handles the website, the database, and the heavy lifting jobs, such as object detection and training,” Hadrien said. “Nectar has GPU functionality, and that’s really handy because we can just reserve it, connect to the virtual machine and temporarily access a lot more processing power when we have images to review. So that’s very useful.”
The platform is also supported by other National Collaborative Research Infrastructure Strategy (NCRIS) facilities, including the Pawsey Supercomputing Research Centre and Astronomy Australia Ltd.
“We use Pawsey’s Acacia hot disk storage system to store the drone images, and we were able to develop the web application thanks to Astronomy Australia’s Astronomy Data and Computing Services (ADACS).”

Finding the Needle in the Haystack
Almost a week since the DFN researchers ventured into WA’s Central Desert, morale was beginning to waver. By that point, they’d searched hundreds of candidates, but were only met with gumnuts, animal droppings and disappointment.
A day that started with 728 candidates was rapidly nearing its end, with only a few dozen candidates left to search. The odds weren’t looking promising.
That was, until they reached the second-to-last candidate, number 727, and one of the researchers radioed in the 6 words that made it all worthwhile: “Guys, I found the fricking rock!”
They found the meteorite.

Despite the joy and relief, Hadrien said that finding it always comes as a surprise.
“As scientists, we do things not because it’s easy, but because it’s hard. We fail a lot at finding meteorites, so when you finally find it after nearly losing hope – that’s the best part,” he said. “Every time we talk about what keeps us going within our team and what motivates us, it’s always going out in the field and seeing our students succeed like they did in this example.”
Three days later, the meteorite was welcomed into its new home in the laboratory at Curtin University, where researchers can analyse it to understand its age, where it came from, and what minerals it’s made up of.
“Meteorites are interesting because they’re really old. They formed right at the beginning of the solar system,” said Hadrien. “So, when you date the formation of these meteorites, you can basically find the age of the solar system. To have something that’s 4.5 billion years old in our hands is quite exciting.”

The success of using drones to search for meteorites has even prompted the team to reopen ‘cold cases’ from the past 10 years – meteorites that would have previously taken too much manpower to search for manually. This includes rare samples, such as potential meteorites from Mars, which would cost space agencies billions of dollars if they were to retrieve them from the Red Planet directly.
Thanks to the hard work and persistence of the DFN’s meteorite hunters, and the ARDC and NCRIS-supported research infrastructure that supports them, what began as a brief streak of light over the Western Australian outback has now become a lasting scientific legacy that illuminates the origins of our solar system.
Find out more about the ARDC Nectar Research Cloud and how to reserve GPUs via the National GPU Service.
The ARDC, Pawsey Supercomputing Research Centre and Astronomy Australia Ltd. are enabled by the National Collaborative Research Infrastructure Strategy (NCRIS).
Written by Dr Cintya Dharmayanti, Scientell. Reviewed by Dr Hadrien Devillepoix (Curtin University), Jo Savill (ARDC), Ben Chiu (ARDC).
