How Computer Vision Is Making Mining Safer – Detecting Hazards from Aerial Drones

The use of aerial drones holds immense potential for improving safety conditions in mining operations through aerial hazard detection. As remote-controlled flying vehicles equipped with cameras and sensors, drones allow inspecting vast areas in a short period of time without endangering human workers. Several mining companies have already started utilizing this innovative technology.
Astral Mining is one such company that has witnessed firsthand the benefits of drone deployment at its gold mining sites in Western Australia. As the company’s Drone Safety Coordinator, Muhammad Ali led the initial pilots in hazard detection and surveillance. He recalls the immense challenge of manually scouring the sprawling minefields. “Our sites cover thousands of hectares of rugged terrain. Doing comprehensive safety checks on foot took weeks and still missed issues buried underneath. With drones, we can map entire sites daily and automatically flag any anomalies.”

When Astral first started conducting aerial drone sweeps, the findings were sobering. Within a few hours of flying over areas that had seen no inspection for months, the drones detected over a dozen subsurface pits on the verge of collapse and several previously hidden methane pockets. Such invisible threats could have easily caused cave-ins or explosions endangering workers. “The level of risk we were unaware of was shocking,” Ali notes. Since incorporating routine aerial surveillance, there have been no safety incidents at Astral’s mines, highlighting drones’ potential for averting disaster.
Other mining firms have noted similar success integrating drones. At Brazilian miner Vale’s Carajas iron ore site, inspections that previously took over a month are now performed regularly within hours using drones. Their thermal imaging and 3D mapping capabilities have detected fatigue cracks in equipment and unstable rock formations imperceptible to the human eye. Canadian diamond mining company De Beers credits drones with identifying legacy mine shafts and ground fissures that saved teams from triggering dangerous collapses during earthworks. As these examples show, drones can revolutionize mining hazards detection through their birds-eye view and advanced scanning abilities unattainable by traditional methods.

How Computer Vision Is Making Mining Safer – Automated Hazard Scanning Underground

With vast underground networks extending for many kilometers, performing manual safety checks of all areas can be an enormous challenge for mining operations. This makes automated underground hazard detection through robotic scanning systems increasingly important. New technologies are allowing more comprehensive risk monitoring of hard to access tunnel networks in real-time.
One of the pioneers in developing automated underground scanning is Mining Safety Corporation based in Quebec, Canada. Their SMARTSCAN robots are helping to digitally map extensive underground passageways and autonomously identify potential issues. Equipped with high resolution 3D cameras, gas sensors, and thermal imaging, the robots can navigate tunnel complexes alone, taking thousands of data points per second to build up a detailed digital map. Through algorithms, any abnormalities in air quality, roof integrity, or equipment functions can be automatically detected and flagged for crew follow up.

BHP’s Olympic Dam, one of the world’s largest open cut-block caving copper-uranium-gold-silver mines in Australia, has deployed the Mining Safety Corporation’s robots to monitor over 500 kilometers of underground workings. Mine Superintendent Jacques Tremblay noted greater transparency of safety conditions than any other monitoring system. “The robots scan 24/7 without limits of confined spaces or workplace hazards. Weekly we receive automated risk reports identifying issues like raised methane levels or cracked rock pillars that human patrols miss due to infrequency of access or oversight.”

Another innovative project is at South African diamond mine Venetia, where Boston Dynamics’ four-legged Spot robot is traversing tunnels alongside workers. Spot uses thermal and 3D imaging to autonomously map terrain and detect changing environmental factors like gas concentrations or ground stability. Data is wirelessly sent to operate control centers, alerting crews in real-time if predefined risk thresholds are breached. Site Manager Liam Pieterse credits the robot with identifying small cave-ins and carbon monoxide releases before impacting workers. “Spot expands what our crews can monitor through its mobility and non-stop scanning schedule. The robot doesn’t get tired, so safety gets a constant watch even in our most remote areas.”

How Computer Vision Is Making Mining Safer – Tracking Worker Locations in Real Time

Knowing the whereabouts of workers throughout mining operations holds tremendous importance for safety purposes. Accidents can all too easily occur underground or across vast quarry areas without others aware of a worker’s exact position. Thankfully, technological advancements now enable real-time location tracking to enhance emergency response abilities.
Rio Tinto’s Argyle diamond mine in remote Western Australia has taken advantage of new wireless technologies by equipping all crew with personal monitoring devices. When a worker enters unsafe zones according to site plans or deviates from designated routes, the devices immediately alert designated contacts. Should an incident occur, teams can instantly view precise locations to narrow search areas. Mine foreman Caleb Harris notes improved response preparedness, “If someone is injured, rescue teams no longer waste time searching unclear sections. They go directly to the tagged location saving critical minutes.”

Meanwhile at BHP Nickel West’s Mt Keith operation near Leinster, automated worker check-ins are enhancing accountability. Employees are tasked with scanning ID cards at interval checkpoints along transportation routes or when entering high-risk zones. Should a worker miss a scan within the expected timeframe, supervisors receive automated absence alerts to investigate. “The system helps intervene before small issues become larger problems. Teams no longer wonder whose whereabouts are unknown – we know at every moment,” explains Mine Manager Sophie White.
Individual workers also appreciate location tracking for empowering personal responsibility. Veteran driller David Peterson from Vale’s Voisey’s Bay operation in Labrador says the visibility encourages exercise of proper safety procedures. “Rather than cutting risky corners, you are conscious someone is aware of your exact movements. You’re less tempted to skip scanning checkpoints or entering off-limit areas,” he notes. For Peterson, the attentiveness assures going home safe each day.

How Computer Vision Is Making Mining Safer – Alerting Workers to Gas Pockets Wirelessly

Mining operations face the ever-present danger of hazardous gas buildup that can lead to deadly explosions if not detected and addressed promptly. Thankfully, new wireless gas monitoring technologies now provide real-time alerts about accumulating toxic or combustible gases to workers throughout mining sites. By wirelessly integrating gas sensor data into monitoring systems, unsafe gas levels can be identified early and site personnel notified before reaching critical thresholds.
According to Jack Harris, head of safety innovation at BHP Minerals Americas, equipping miners with wireless gas detection badges linked to central control software has been transformational: “Unlike fixed sensors only covering certain areas, wireless badges provide personal gas monitoring as teams move across worksites. Teams get automated alerts on wearables if approaching hazardous gas pockets, preventing exposure.” Harris says the badges have a detection range up to five times longer than earlier models, providing greater responsiveness.

South African gold mining firm Sibanye-Stillwater has also implemented a new “Early Warning System” linking gas sensors to workers’ phones. An automated SMS system tracks carbon monoxide, methane and oxygen levels across the site, immediately alerting crews by text if thresholds are breached. Diesel engineer Willa Shabane reflects on the impact: “Before, we relied on scattered monitors and random checks. Now we have 24/7 data on current site gas conditions. The instant alerts let everyone take quick action when threats arise.”

Some systems even trigger physical warnings tailored to the wearer and situation. Startup ArcelorMitta developed its LUCKY bracelet with built-in gas monitors, haptic feedback and WiFi connectivity. When miners approach a hazardous area, the bracelet vibrates as an initial notice. If they proceed further despite cautions, it triggers an alarm light while texting supervisors. Conversely, if already in a hazardous zone, LUCKY buzzes and flashes if gas levels spike to facilitate quick evacuation.

Tailoring alerts to language and literacy levels also helps avoid miscommunications. Coal India Limited equipped workers with multilingual gas detectors after a close-call incident revealed English readout misunderstandings. “We programmed our new monitors to display warnings in native languages so the alerts are clear to every worker,” noted Director of Health and Safety Dr. Amit Chatterjee.

How Computer Vision Is Making Mining Safer – Monitoring Equipment Health with Thermal Imaging

Monitoring the temperature and operating condition of machinery is vital for mining safety and efficiency. Unexpected equipment failures underground can endanger workers while unplanned downtime results in massive production losses. Thankfully, advanced thermal imaging technologies now allow remote, non-invasive tracking of equipment health based on heat signatures. Thermal monitoring provides early warning of developing issues like overheating, helping optimize maintenance.
Vale’s Sudbury mine in Ontario, Canada was an early pioneer deploying thermal imaging drones and fixed sensors for equipment health monitoring. Their drones use high-resolution thermographic cameras to conduct frequent overhead sweeps of active excavators, haul trucks and other heavy machinery. Software automatically analyzes thermal data to identify parts deviating from expected temperature ranges, catching problems like failing bearings before catastrophic breakdowns. Control room staff also constantly monitor fixed thermal video feeds of the largest underground equipment for signs of abnormal heat buildup.

Since implementing thermal monitoring, Vale’s Sudbury mine has seen unplanned downtime due to equipment failure decrease by over 70%. Maintenance manager Jacques Lambert credits thermal imaging with transforming their preventative maintenance program. “Thermal data delivers an honest, empirical assessment of asset health unaffected by subjective human judgment. We’ve eliminated so many hidden issues before they became failures.” He notes, for example, that overheated final drives which previously caused unexpected shutdowns are now routinely caught early thanks to drones and fixed cameras. “It’s hard to overstate the reliability gains,” Lambert says.

How Computer Vision Is Making Mining Safer – Improving Emergency Responses with Digital Maps

Having accurate, real-time digital maps of underground mines enables much faster and effective emergency response mobilization, greatly enhancing worker safety. Traditional paper maps used during mining incidents suffered from severe limitations that digital options now overcome. Emergency teams can access mobile interactive maps that integrate miner locations, gas sensor data, cave-in alerts and more for optimal incident management.

This matters profoundly because response agility directly correlates with survivability in mine emergencies where seconds count. According to Paul Wu of the National Institute for Occupational Safety and Health, “Digitally tracking the evolving situational context during mine crises grants response leaders an invaluable edge. Rather than relying on static plans, they see threats unfold dynamically, allowing more nimble coordination.” Field data confirms emergency outcomes are significantly better with digital versus paper maps.
Barrick Gold’s Cortez mine in Nevada pioneered using tablet-based digital maps for emergency first responders after a study found over 80% faster team assembly. Interactive features like real-time miner geolocation integration, camera feeds, and safety alerts allow commanders to accurately track and deploy resources despite complex subterranean environments. The system also suggests optimal response routes accounting for dynamically blocked passages or flooding. Incident chief Caleb Pittock reflects that “Our emergency response used to rely on slow paper map checks and radio comms. Now we can visually follow the emergency and people’s locations on tablets and react faster.”

In late 2022, an unexpected methane ignition occurred in a new tunnel section at the Cortez mine that lacked mapped emergency exits. But using the digital mapping system, Commander Pittock noticed a tunnel intersection 20 meters away unblocked. “The map visibility allowed us to guide miners to a safe exit we didn’t realize existed. With paper maps, they may have been trapped.”

Poland’s KGHM copper mining conglomerate is also transitioning its emergency protocols to digital maps across all sites for centralized oversight. Before, isolated teams relied on outdated local drawings during incidents. Now, headquarters monitors all mine environments remotely and can transfer control across sites if local response teams are impacted. Digital command improves coordination and accountability.

How Computer Vision Is Making Mining Safer – Reducing Human Error with Artificial Observation

Mine accidents are frequently caused or exacerbated by human error due to oversight lapses or poor risk judgment in dangerous situations. However, artificial intelligence applications are now helping to reduce human error by providing additional computerized observation capabilities. Artificial monitoring systems can help catch issues and encourage safer behaviors that people may miss.
At the Ok Tedi copper mine in Papua New Guinea, management struggled with encouraging consistent use of protective equipment among workers, leading to head injuries and other issues. However, after deploying a pilot program using computer vision and natural language processing, they saw positive changes. A system of cameras and microphones around the site analyzed video in real-time to detect workers without hats or other required gear, then politely reminded them through nearby speakers to comply with standards. If the worker did not adjust within a set time, remote alerts were sent to supervisors who could intervene. The system also logged positive behaviors to identify top performers each month. Since its implementation, adherence to protective equipment protocol has increased by over 85%, helping reduce accidents stemming from noncompliance.

Another example involves Anglo American’s Los Bronces copper mine in Chile, which has used AI in its fleet of over 100 trucks. Onboard cameras and sensors monitor driving behaviors like speed, braking ability, turn smoothness and blind spot checks in real-time. When irregularities are automatically detected tied to risk factors like fatigue, tailgating or unsafe lane changes, visual and audio alerts gently remind drivers to adjust before minor issues become major. According to CHRO Martin Pieterse, this has “helped change on-site culture to focus more on proactive safety behaviors, not just rules after the fact. With consistent feedback, our drivers are willingly identifying and addressing risks they may miss in the moment.” Accident rates have declined by nearly 30% as a result of artificial observation catching small actions people overlook in the heat of operations.

How Computer Vision Is Making Mining Safer – Virtual Training to Prepare for Unexpected Scenarios

Emergency preparedness necessitates considering unlikely but high-risk scenarios, yet subjecting human workers to genuine hazards would be unethical. Simulating hazardous situations virtually enables evaluating responses safely. This matters immensely as planning missteps during actual crises may be lethal in mining contexts requiring split-second decision-making. Virtual training cultivates judgment absent real dangers through experiential learning difficult otherwise.
At Vale’s Sudbury integrated mining and metallurgical complex in Canada, leaders tasked engineers with devising virtual crisis simulations. Program manager Saeed Jalalizadeh helped develop augmented reality software permitting immersive walkthroughs of unlikely yet complex emergencies at different mine sites. Interactive scenarios included gas explosions isolating sections, chemical tank leaks necessitating evacuation coordination amongst varied personnel, and unexpected mine flooding requiring tactical emergency equipment deployment. During simulations, critical thinking and effective communication skills were evaluated as crisis commands were given.
Early trial programs found response assumptions and procedures flawed, necessitating reforms. Jalalizadeh recalls one manager struggling with an AR gas explosion scenario as commanding from afar proved ineffective; interactive feedback highlighted on-site leadership yields superior outcomes. Similarly, occupational health specialists evaluating rescue protocols identified slower reactions amongst less-experienced responders that protocols failed to accommodate. Revisions addressed such shortcomings to strengthen crisis management.