Drive-through ore sorting and other mining automations
Posted: October 17, 2025
To the layman, sorting ore seems quite straightforward. You dig up a rock and, if it contains what you’re looking for, extract the metals or minerals. If not, discard and try again.
Of course there’s a lot more to it in practice. But as it turns out, the process has only become more hit-or-miss in the past few decades.
That’s because, after centuries of mining the richest deposits, ore grades for key commodities like copper, iron and gold are declining in many places. This means more money and energy have to be spent on extracting the metals, and more waste is produced in the process.
Enter NextOre, a mining technology start-up founded in partnership with Australia's national science agency CSIRO. The company has spent years developing a system to make the process of sorting copper from waste more cost-effective, drawing inspiration from an unlikely place: hospitals.
Our Industrial Life
Get your bi-weekly newsletter sharing fresh perspectives on complicated issues, new technology, and open questions shaping our industrial world.
How NextOre built a giant MRI scanner for mines
Copper is perhaps the most essential metal for the energy transition, found in generators, turbines, solar panels, batteries and the electrical grid itself. Experts have long warned that we’re headed for a supply crunch, however: Demand for the metal is set to outstrip supply within the next decade, according to the International Energy Agency.[1]
Aside from tapping new deposits—which can take well over a decade—McKinsey says new technologies to extract more copper in the processing plant can help bridge the supply gap.[2]
NextOre aims to optimize copper production even earlier in the processing chain, by more selectively sorting the ore before it even leaves the mine. Its advanced sensor system uses magnetic resonance technology—the same as an MRI scan at the hospital—to send short pulses of radio waves through collected ore. The company says the technology can measure grades much faster and more accurately than conventional sensors.
The first commercial version of its MRI scanner, launched in 2018, was installed along conveyor belts and sent to several large producers for trials. It’s now sorting ore at several sites, including one plant in Chile scanning 6,500 tons per hour.
Now, after several years of additional research and development, a new version scans ore trucks before they ever leave underground mines. Trucks drive through a shed with a seven-meter-wide disc-shaped scanner suspended from the ceiling, which can analyze a load in less than a minute, depending on the quality (lower-grade ores take longer to scan).
The benefits are obvious: Sorting ore close to where it’s mined, and only sending the highest-quality material to a processing plant, saves time, resources and waste further down the line.
Integrating robotics into NextOre’s underground scanner
Developing the new system was more involved than simply mounting the MRI scanner in a shed, however. After fine-tuning its technology together with the scientists at CSIRO, NextOre “soon realized that extensive automation, controls and programming were required to implement the technology on mine sites,” says Shivika Singh, the firm’s Chief Operating Officer.
To work out how the scanner could operate in the field, the company turned to the University of Technology Sydney’s Robotics Institute. There, a team around Associate Professor Gavin Paul went to work figuring out how to position the sensor to maximize coverage and get the most accurate readings. The researchers and students at UTS were also tasked with automating the communication between large cranes and heavy vehicles to avoid collisions in tight underground mines.
“We integrated CSIRO’s lidar sensors into our prototype to scan load volumes, identifying the highest points and contours,” explains Paul. “We computed the optimal sensor placement, positioning one at the front, one back of the load and another tracking the truck as it approached.”
On top of the robot vision, the team also added a command system to automatically move the crane holding the scanner into position above the truck—eliminating the need for a human operator to calibrate the placement for each load.
The final prototype of the new system was tested at First Quantum’s Kansanshi copper mine in Zambia, the highest capacity ore sorting plant in the world. NextOre says the unit will be used for up to 50-tonne mine trucks initially, with plans to adapt it for underground bucket loaders in the future. And while applications have so far focused on copper, the company says the MRI also works on other ores, such as lithium and iron.
Managing data in an automated mining industry
While NextOre seeks to master ore sorting, miners have also turned to a slew of other innovations to automate and optimize their operations.
BHP started using autonomous drills at its iron ore mines in Western Australia in 2016 and, by 2022, had already expanded its fleet to 26 rigs across five mine sites. Its drills, in the dry Pilbara region, are operated remotely from Perth, hundreds of miles away. Last year, the company finished converting all 33 trucks and 5 drilling rigs at its Spence copper mine in Chile to full automation—a likely blueprint for other operators chasing efficiency.
Other major mining companies, such as Anglo American and Rio Tinto, have similarly invested in autonomous drilling, haulage and bulk ore sorting. And on top of automating existing equipment, miners now have access to new tools like robots that can navigate and map flooded mines, and drones for conducting aerial surveys, monitoring and inspection.
“Drones can go where it is risky for people, like pit walls or tailings dams, or into hazardous conditions,” says Rebecca Kahrhoff, industry manager for mining at geographic information system company Esri. “What used to take days can now be done in a few hours and on an as-needed basis—and you get better, high-resolution, real-time data that helps you enhance decision-making.”
One thing that is getting harder as mining increasingly automates: managing all that data.
Although AI and machine learning are widely used across the industry already, experts at Australia’s CSIRO say the sheer breadth of data inputs is challenging. “Full automation of mining operations relies on vast amounts of highly diverse data—and this can present a unique problem for miners,” says Ewan Sellers, formerly the agency’s Digital Mining Lead and now a manager at BHP.
The shift is already changing the most in-demand skills in the sector, towards mechatronic engineers, electrical technicians and software engineers. Mining companies also have to train more staff to be able to supervise their machines.
In other words, today’s miners are less likely to descend into the pit and more likely to recline in wired-up office chairs fitted with joysticks and facing a wall of monitors—a long way from the shovels and pickaxes of old.
