Every person who works in underground mining knows the routine. Before production resumes after a blast, someone has to go in and inspect the heading. Check the ground conditions. Look for loose rock on the backs and walls. Assess the ventilation. Confirm it’s safe for the crew to return.

It’s critical work, and it’s inherently dangerous. The person doing the inspection is, by definition, entering an area that hasn’t yet been confirmed as safe. They’re the canary. And despite decades of improvements in ground support, scaling practices, and geotechnical monitoring, rockfall in underground mines remains one of the leading causes of serious injuries and fatalities worldwide.

Drones are beginning to change this. Not as a novelty, not as a proof of concept, but as a genuine operational tool that’s keeping people out of harm’s way.

The Risk Profile of Underground Inspections

The statistics paint a sobering picture. According to data compiled by the International Council on Mining and Metals, ground control failures — which include rockfall, ground collapse, and seismic events — account for approximately 30-40% of underground mining fatalities globally, depending on the year and region.

In Australia, the record is better than the global average, but it’s far from zero. Every state mining regulator publishes incident reports that include near-misses from rockfall during inspection activities. These near-misses don’t make the news, but they represent situations where, with slightly different timing or positioning, someone would have been seriously injured or killed.

The vulnerability window is clear. Post-blast inspections, stope void assessments, and ventilation checks in abandoned or partially collapsed areas put people in locations where the ground conditions are uncertain. That uncertainty is the entire reason for the inspection — if you knew the ground was stable, you wouldn’t need to check. But checking requires exposure.

This is the fundamental paradox that drones address.

What Underground Drones Can Actually Do

Underground mine drones in 2026 are a different beast from the consumer quadcopters that people picture. The operating environment is brutal — no GPS, limited communication, dust, humidity, explosive atmospheres, and darkness. Consumer drones are useless in this context.

Purpose-built underground drones — from companies like Emesent, Exyn Technologies, and Flyability — are designed around these constraints. Key capabilities include:

LiDAR-based autonomous navigation. Without GPS, underground drones navigate using simultaneous localization and mapping (SLAM) algorithms built on LiDAR sensor data. The drone builds a 3D map of its environment in real time and navigates using that map. This allows it to fly into spaces it’s never entered before — exactly the capability needed for post-blast inspections.

High-resolution visual and thermal imaging. Cameras capture detailed imagery of backs, walls, and faces that geotechnical engineers can review remotely. Thermal cameras can identify areas of water ingress, heat buildup, or other anomalies that indicate potential ground instability.

Gas detection sensors. Multi-gas detectors mounted on the drone measure concentrations of methane, carbon monoxide, nitrogen dioxide, and oxygen levels. This is critical for post-blast re-entry assessments and for evaluating areas where ventilation may be compromised.

3D point cloud generation. The LiDAR data produces dense 3D models of the inspected area, which can be compared against the designed excavation profile. This identifies overbreak, underbreak, convergence, and other deviations that indicate ground stress or support deficiency.

Real-World Deployment: What’s Working

Emesent, an Australian company that spun out of CSIRO, has become one of the leading providers of underground drone technology. Their Hovermap system has been deployed at operations across Australia, Canada, and South America.

At a gold mine in Western Australia, Hovermap drones are now conducting routine post-blast inspections of development headings. The previous process involved a geotechnical engineer and a mines inspector physically entering the heading, spending 15-30 minutes assessing conditions, and then signing off for production to resume. The drone completes the same assessment — visual inspection, ground condition mapping, atmospheric testing — in about 10 minutes, with no human exposure.

The time savings are meaningful, but the safety improvement is the real story. Over a twelve-month trial, the operation identified six inspection events where the drone detected conditions that would have posed a direct risk to a human inspector — loose rock on the backs that wasn’t visible from floor level, gas concentrations above threshold, and in one case, active ground movement that was captured in real time by the LiDAR.

Six events in a year. At a single operation. That’s six potential injuries or fatalities prevented by keeping a person out of the space.

At an underground copper mine in South Australia, drones are being used for stope void surveys — mapping the shape and condition of large open voids that are too dangerous for human entry. Previously, these surveys relied on cavity monitoring systems (CMS) that gave limited data, or they simply weren’t done because the risk was too high. The drone-generated 3D models provide information that was previously unavailable, improving stope design and reducing dilution.

The Limitations Nobody Should Ignore

Underground drone technology isn’t perfect, and anyone selling it as a complete replacement for human inspection is overstating the case.

Flight endurance is limited. Most underground drones can operate for 15-30 minutes per flight. In a large underground operation with kilometres of development, that means multiple flights and battery swaps to cover a significant area. You’re not replacing a shift-long inspection walk with a single drone mission.

Environmental constraints are real. Heavy dust after blasting can overwhelm LiDAR sensors and cameras. High humidity causes condensation on lenses. Strong ventilation currents can make stable flight difficult in larger openings. Operators report that about 15-20% of planned drone flights are delayed or modified due to environmental conditions.

The human element still matters. An experienced geotechnical engineer walking through a heading picks up on subtle cues that current sensor technology can’t reliably detect — the sound of stressed rock, the feel of loose material underfoot, the smell of fresh ground water. Drones capture a tremendous amount of data, but they don’t replicate the full sensory toolkit that humans bring to an inspection.

Data processing takes time. A single drone flight can generate gigabytes of point cloud and imagery data. Processing this into actionable information — geotechnical assessments, convergence measurements, stability classifications — requires skilled personnel and time. In some operations, the bottleneck isn’t the inspection itself but the analysis afterwards.

The Workforce Conversation

There’s understandable concern among mine workers about drones replacing jobs. It’s worth addressing directly.

The roles most affected aren’t being eliminated — they’re being moved out of harm’s way. The geotechnical engineer who previously walked into the heading now reviews the drone data from a control room. They’re still making the assessment. They’re still applying their expertise. They’re just not doing it while standing under potentially unstable ground.

If anything, the demand for skilled geotechnical and inspection personnel is increasing, not decreasing, because drones generate far more data per inspection than a human walkthrough. Someone needs to interpret that data, and it requires the same geological and engineering knowledge it always has.

The shift is from physical exposure to information processing. For a workforce that’s long accepted dangerous conditions as part of the job, that’s unambiguously positive.

Where This Goes Next

The next generation of underground drones will likely integrate more advanced AI — automated detection of geological hazards, real-time comparison against design models, and predictive alerts based on observed ground behaviour patterns. Some of this is already in development.

Multi-drone operations — swarms that can map an entire mine level simultaneously — are technically feasible but still a few years from operational deployment. Communications infrastructure underground remains a constraint, though mesh networking and 5G trials in mines are progressing.

The ultimate destination is an underground mine where no person enters an area that hasn’t been robotically assessed first. We’re not there yet. But the gap between where we are and where we’re going is narrowing faster than most people in the industry expected.

Every miner who goes home safe at the end of a shift is a success. If drones can improve those odds even slightly, the investment case makes itself.