From drill core to quantum leaps CSIRO’s sensors are turning rock whispers into real-time revelations that miners can actually use

Dr Sandra Occhipinti, research director in minerals at Australia’s national science agency, CSIRO, is leading a team of more than 100 scientists focused on one of the most complex challenges in modern exploration: how to accelerate mineral discovery in covered terrains while simultaneously improving geometallurgical insight across the mining value chain.

With a career spanning academia, government, and industry, Sandra brings a pragmatic yet forward-looking approach to technology deployment. Her remit is the Discovery Program, a major research initiative within CSIRO aimed at delivering new ways to locate and characterise mineral systems in Australian conditions.

“We’re tasked with figuring out new ways to find ore bodies, but to do that, we have to understand how these systems form through geological time,” said Sandra. “Often, you’re not detecting deposits directly; you’re detecting subtle clues in the rocks. So we build tools that help industry recognise and act on those signals in real time.”

Defining real-time ore characterisation

According to Sandra, CSIRO defines “real-time ore characterisation” as the combination of rapid data collection and immediate interpretation.

“There’s no shortage of sensors that can scan core or faces rapidly, ” she explained. “But the real value comes when that sensing is combined with on-the-spot mineralogical or lithological interpretation. That’s what turns data into decisions. ”

This distinction is critical. Real-time characterisation is not just about speed; it’s about insight delivered fast enough to influence decisions in the field, in the pit, or in the plant.

From HyLogger® to Maia Mapper: Tools reshaping ore understanding

Two standout tools have had an outsized impact on how industry understands and models ore systems: HyLogger® and Maia Mapper.

“HyLogger started in the early 2000s and is now up to version four, ” said Sandra. “It scans drill core for mineralogy, which is essential because minerals host the metals we’re chasing. ”

HyLogger enables geologists to understand their ore systems faster, without waiting weeks for lab assays. It has also become a key part of the National Virtual Core Library through AuScope, creating the world’s largest database of drill core with associated spectral data.

“We just take it for granted here, but there are over 6,000 drill cores in that database. No other country has anything like it.”

Maia Mapper, meanwhile, is a high-resolution scanning instrument used in CSIRO labs. It provides petrogenetic insights by generating spatial geochemistry maps from large core sections.

“It’s not a field sensor, but it brings synchrotron-level capability into labs that industry can access,” she said.

Most transformative points in the value chain

Sandra said the point of greatest impact for real-time sensing depends on the deposit type and operational priorities.

“We see these technologies used right along the value chain - from early exploration and drill-outs to pre-concentration and processing,” she explained.

Applications range from brownfield targeting and geotechnical modelling to pre-concentration screening and deleterious mineral detection.

“In processing, for example, you want to avoid putting waste rock through the plant. Real-time tools can help sort material before it reaches that stage.”

A smarter alternative to lab-based workflows

CSIRO-developed and supported platforms offer significant advantages over conventional lab workflows.

“One of our key advantages is speed,” Sandra said. “We’re building systems that give you usable mineralogy in the field. We’re also embedding domain expertise directly into the toolsets.”

In some cases, CSIRO also modifies commercial instruments with bespoke software. A prime example is the LIBS-based system CSIRO developed to map mineralogy on core. The software behind it is trained on millions of SEM spectra.

“That’s what makes it robust. We’re not just bolting on AI for the sake of it - it’s backed by decades of geoscientific data.”

Integrating sensors with geoscientific modelling

One of the hardest problems CSIRO tackles is integrating disparate datasets from sensors into broader geoscientific models, geometallurgical workflows, and decision support tools.

“Different instruments collect data at different scales and resolutions,” said Sandra. “We’re working on ways to harmonise those inputs.”

This includes building data systems that allow for scalable integration and running controlled experiments to compare sensor outputs across platforms. The result: better, more coherent inputs for 3D models and geomet decision-making.

CSIRO’s Exploration Toolkit (XT) platform helps by housing software and algorithms in one place for staff and industry. A recent addition is LandScape+, which helps users map landscape domains to normalise geochem data.

Where machine learning and AI fit in

Sandra said AI plays a major role in simplifying interpretation of large datasets generated by tools like Maia Mapper.

“We used to rely on GeoPixie, a powerful but complex tool. Now we’re developing Maia Cube, which applies machine learning to translate the same data into usable mineralogy much faster.”

CSIRO also employs machine learning in LIBS workflows, where models trained on historical SEM data are being used to automate mineralogical interpretations.

“AI is only as good as the training data. We’re very focused on ensuring that our models are transparent, well validated, and not just black boxes.”

Evidence of success in the field

While companies don’t always publicise results, Sandra shared that CSIRO sensing technologies have driven operational gains.

“We know that HyLogger-like technology combined with The Spectral Geologist software has been used to upgrade ore and optimise processing. We also understand that that process has been copied in various forms, which is great and tells you it works.”

Another success story is LIBS technology. CSIRO has worked with Northern Star for over a decade, starting with SEM-based mineralogy and pivoting to LIBS as the technology matured.

“It’s been very successful for them. We can’t share all the details, but it’s clear the tech is delivering value.”

Barriers to broader adoption

One of the biggest challenges is trust - particularly around the interpretation of sensor data.

“We need to do a better job of explaining what the sensors are measuring, what assumptions are baked into the models, and what the limitations are,” said Sandra.

She also cautioned against overpromising. “If we’re not clear about what the tech can and can’t do, users won’t trust it.”

Scaling through partnerships

CSIRO scales its tech through commercial and research partnerships. In many cases, it licenses innovations to third parties that can take the tech to market.

Examples include CoreScan (HyLogger) and Portable PPB (DetectOre). In both cases, CSIRO provided the science and industry partners provided the commercial traction.

“We don’t want to be the commercialisation engine. We want to stay focused on science and let others scale the solutions.”

The road ahead: Quantum, fusion, and trustable AI

Looking forward, Sandra sees major opportunity in quantum sensors, quantum algorithms, and large-scale data fusion.

“Imagine 3D models with built-in uncertainty, run thousands of times quickly and cheaply to find the most suitable scenario. Quantum tech could make that a reality.”

CSIRO is also working on ways to fuse data from satellites, drones, and drill core to support faster, more robust exploration and processing decisions.

“There’s so much knowledge already in CSIRO. It’s not a giant leap to get to the next level - just a well-informed step.”

Mining’s digital frontier

With mounting pressure to discover deeper, process leaner, and reduce environmental impact, CSIRO’s science-backed, application-driven approach to real-time characterisation is emerging as a cornerstone of the mining industry’s digital future.

“We don’t have all the answers,” Sandra concluded. “But we’re building the tools to help you find yours."