By Steven Micklethwaite, Sustainable Minerals Institute, and Friederike Körting, HySpex by Norsk Elektro Optikk.
Drones have been firmly embedded in our imagination, especially thanks to the world of cinema and television. Society and industry experts are becoming increasingly inventive in how we can benefit from these robotic aircraft; enter the minerals industry.
Drones have become a familiar feature of mine sites, with rapid incorporation in some aspects of operations, to the point where they are a routine tool for stockpile surveys, some equipment inspections and tailings inspections. More famously, the extractives industry has made the tunnel-mapping drones imagined in movies like Prometheus and The Incredibles a reality, through Emesent’s ‘Hovermap’ device.
Globally, the standard workhorse of drone sensing has been photogrammetry. Attach an affordable camera on a drone, combined with some decent GPS, and with computer vision magic we suddenly have accurate photorealistic three-dimensional models of our sites. But there is a dazzling and growing array of different sensors or geophysical instruments that can also be mounted.
Sensing technology adaptation
The first place that both the minerals industry and the drone sector have gone is to adapt sensing technology that already exists in some form, on aircraft or satellites. The world’s largest commercial manufacturer of drones, DJI, has for a few years now, sold platforms integrated with thermal and multispectral cameras for enterprise applications.
Multispectral cameras capture light in specific wavelength ranges, or ‘bands’, from the visible and near-infrared parts of the electromagnetic spectrum (400-1000nm). They effectively capture a dimension of the world that our eyes are incapable of seeing. We can use these bands to make calculations, such as different types of ratios, that reveal everything from vegetation cover to deep water structures, soil disturbance and much more.
At the University of Queensland’s Sustainable Minerals Institute, we have used multispectral cameras to help track rehabilitation, monitor mine closure sites and map out the distribution of iron oxides in soils and rocks as a proxy for acid mine drainage contamination.
As such, with some straightforward and relatively cost-effective equipment, we can use the same drone platform to provide valuable information for different aspects of operations. In other words, drones are interoperable. In this case they provide opportunities for multispectral and photogrammetric technology and resulting insights to be shared between the survey, tailings, environmental and closure teams.
Nonetheless, we know we can do better. Multispectral cameras catch only limited information in their ‘bands’ and have significant gaps between the bands. The most common sensors use the visible and near-infrared parts of the spectrum but miss the short-wave infrared. If, however, our cameras can collect data continuously from the visible all the way to the short-wave infrared (400-2500 nm wavelengths), in narrow increments rather than broad bands, then we open a new world of possibilities.
Making the invisible visible
This is where hyperspectral cameras come in, otherwise known as hyperspectral imaging.
Hyperspectral cameras are not new and have been sensing our world from aircraft, tripods, satellites or mounted to conveyor belts for some time. Due to their weight however, short-wave hyperspectral cameras for drones are new.
We are now beginning to optimise these cameras and overcome the weight barrier. In fact, the next-gen hyperspectral instruments for drones are able to span both the visible near-infrared to short-wave infrared (VNIR-SWIR) spectrum all in one.
Research has shown that in this era of ‘beyond visible’, there are many new avenues for application, especially with short-wave infrared at our disposal. In particular, being able to collect high-resolution VNIR-SWIR data helps map individual minerals of key significance to mining, such as carbonates, sulphates and sulfosalts, clays, the different iron oxide species and the phyllosilicate minerals such as chlorite, talc, muscovite and others.
This bewildering array of minerals is well-known in the world of exploration as important for vectoring towards resources. However, the minerals have perhaps even more value for geometallurgy and mineral processing, opening the possibility to use drones as an early warning system from bench-to-mill, to track the ore and waste entering mineral processing plants.
Other applications include monitoring contaminated water chemistry along streams, or mapping varieties of plant species and their health in areas of rehabilitation. Hyperspectral cameras mounted on drones have already been used to create automatic geological maps of open pits, where no geologist is allowed to access.
Satellites have had shortwave infrared, and indeed other useful wavelengths, for some time. The main problems are the obstructions caused by cloud cover and the spatial resolution of their individual pixels being in the order of metres to tens of metres. A single pixel resolution of 10m by 10m is really composed of many different minerals and so it is a challenging exercise to process satellite data, and deconvolute the spectral signal.
Overcoming barriers with drones
The unique characteristics of drones enable us to overcome some of these drawbacks.
For a start, drones can be flown at lower altitudes, close to the ground’s surface. This means they collect data below cloud cover and with very high pixel spatial resolution. In fact, it is possible to get sub-cm resolution if required.
Secondly, drones can do repeat surveys more easily and can potentially scan steep walls as well as large flat areas. This means that we can use drones for change detection and do so in lots of different scenarios. This quality is a good example of where drones allow something new that was not previously possible.
It is not hard to imagine daily flights of production benches and pit walls leading to semi-automatic geological and resource model updates, and much more regular reconciliation. That same instrument can be tracking changes in infrastructure integrity due to corrosion, over major capital investments such as processing plants.
It almost sounds too good to be true. So why is our industry not routinely using this technology or pushing it forward? Simple sound bites and familiar accusations would point to conservatism in the minerals industry, but these accusations are unfair.
The truth is that the technology is challenging. There is already a level of compliance beyond conventional workplace or mining regulations, related to aviation authorities. This is not a barrier for the industry because we are very good at working to regulatory frameworks, but it does require training, education and time commitments from mining professionals who are already stretched.
The main challenge is that the processing and interpretation of hyperspectral data requires deep expertise. The enterprise software solutions for doing this are not easy to drive and, more importantly, it can take up to three weeks to get results. Such turnaround times are fine for the exploration and environmental management arms of the industry but are not suitable for operations and production, where the most immediate value lies.
The M4Mining consortium
Technology is meeting these challenges now, and the Sustainable Minerals Institute and its European partners are providing solutions in a major European project called M4Mining.
This three-year project will deliver a turnkey drone hyperspectral system by HySpex that can return calibrated, corrected and even interpreted data for mining professionals within 24 hours; in fact, the goal is as little as four hours.
It is an exciting project where we will combine the drone with a hyperspectral imager, LiDAR, a machine vision camera, a gimbal that can rotate the camera at multiple angles, and importantly the software modules for the onboard processing. These technical aspects are only half the journey.
One of the objectives of M4Mining is to develop and demonstrate a cost-effective method for regular mapping, monitoring, and geo-hazard prediction, as well as management of mining operations, including stockpiles and tailings over time. This means running a series of case studies across Australia and Europe in discovery, operational, closure and post-closure environments.
We are also developing whitepapers on the operational context of such instruments, and sampling strategies so that industry will be well-positioned for best practise.
A final key component is to understand how drone data can integrate with satellite data to provide a much better product for industry. Both technologies should be viewed as complementary not competitors.
Drones have limited battery life and though they can be extremely cost-effective for certain applications, they bring more value to the industry when we use their data to improve the quality of satellite data which can cover much larger surface areas.
None of this work is occurring in a vacuum. Without infringing IP, some of the data and publications will be made public through the lifespan of the project. This project also remains open for industry participation and we are actively looking for engagement from those partners who wish to gain early insight from the opportunities that exist ‘beyond visible’.
Drone infrastructure at the Sustainable Minerals Institute is enabled by AuScope and the Australian Government via the National Collaborative Research Infrastructure Strategy (NCRIS). To learn more visit auscope.org.au.
M4Mining is funded by the European Union’s Horizon Europe programme under Grant Agreement ID 101091462. For more information, visit www.m4mining.eu