Like the famous canaries that were first brought into coal mines from the early 1900s to detect carbon monoxide, creative solutions have long been used to mitigate some of the risk inherent in mining. Even so, human bodies have long borne the brunt of this cramped and strenuous occupation, from the creeping 'black lung' of coal miners to blasting injuries from explosive flyrock. So it's good news that the new frontier in mining is to remove them from proximity to physical danger as much as possible, perhaps one day entirely.
That's why automation is a big deal in heavy industries like mining and construction. Leading names of the industry, like British-Australian firms Rio Tinto and BHP and Canada's Barrick Gold, are all investing large sums of money in automating more and more of the process of mining, including an increasing use of driverless vehicles. But you can dispel images of retrofitted Teslas from your mind: these vehicles are seven metre high (23ft) giants that can move at 40mph and carry a 230 ton payload.
According to Rio Tinto, these vehicles can boost productivity while helping to improve mine safety at the same time. In the company's Australian mines, for example, many first generation driverless vehicles are actually operated remotely by crew hundreds of kilometers away, taking drivers away from the punishing summer heat (sometimes in excess of 110° F) and cutting down the risk of errors induced by fatigue or environmental stress.
These remotely operated vehicles are already a boon to the industry, but the latest generation of driverless mining vehicles remove the need for a human operator entirely. Made by Komatsu, a Japanese construction equipment company (and in fact the second largest manufacturer of mining equipment in the world behind Caterpillar), the new line of unmanned haulage vehicles was unveiled at MINExpo 2016, the industry's flagship equipment show, where they made waves with a completely cabless design.
But even with all of the sensors we already associate with autonomous vehicles—stereo cameras, GPS, LiDAR, etc.—automation challenges faced by mining machinery are different from those experienced by, say, self-driving cars cruising down Main Street. Mining involves a different level of interaction with raw materials like stone or sand, along with the need to work collaboratively with other humans and machines, all while navigating irregular and uneven paths or confined spaces. In order to do this, accurate modelling of the physical environment is key—and that's where the drones come in.
In the past few years Komatsu has developed a system called Smart Construction, a project that makes use of services from Skycatch, a San Francisco-based startup which specializes in providing enterprise customers with super accurate data collection and analysis using drone photography. Essentially, in the Smart Construction process a small drone flies over a target area in a pre-programmed pattern, taking a series of photos of the ground below. The drone then uploads the photographs to another piece of software, developed by Skycatch, which processes the data to produce a 3D rendered topographical model of the area in centimeter-level detail; with this complete, a foreman can use a computer system to direct robotic vehicles around the site.
Although this software can be used with many different camera/drone combinations, for precision operations Skycatch sells its own drones to clients, which come with a custom made RTK antenna system built into industry leading DJI drone hardware. The precise combination of photography and geospatial data is what allows for the real technical wizardry, as Skycatch founder and CEO Chris Sanz explained.
"The drone automatically connects to an RTK antenna to triangulate the location of the drone itself when taking the photo," said Sanz in a Skype call. "Microsecond synchronisation of where you are in space where the photo is taken is what gives us the data we can use to make a centimeter accurate point-cloud… With the point-cloud you can then make measurements, put it into your AutoCAD or design tools, or you can load it into a bulldozer and automate the movement."
Compared to traditional surveying techniques, drone photogrammetry reduces the time taken to make a comprehensive model of a work site from a matter of weeks to just a few hours. The ability to get an up-to-date snapshot of the site so quickly also makes a new automation workflow possible – which if the promotional video is to be believed, looks a bit like a deploying units on a Command & Conquer game:
As this trend towards automation continues there will almost certainly come a point where humans no longer need to be present in mining areas at all. In the interim period, mines will increasingly see a mixture of autonomous and human-operated machinery, which brings its own challenges. For example, although autonomous vehicles are working according to logical rules, these may not always be apparent to human workers: in one accident recorded in an Australian mine, the route pre-programmed into an autonomous truck had not been physically signposted in the mine, leading to a serious collision when the automated vehicle unexpectedly cut across the path of a vehicle with a human driver. (On some levels this is still a case of human error, but an error that would have been apparent had both vehicles been manned.)
However, setbacks notwithstanding, the path is laid out toward a new paradigm for mining, and one that has implications beyond the extractive industries as automation and robotics start to play a greater role in modern life. Mines are far from the only places that can benefit from the kind of rich data gathered by drones, as Skycatch's Chris Sanz explained.
"We're becoming a platform that allows enterprise companies to have accurate access to the physical world in real time," he said. "One day this is going to change so many different industries that we can't even imagine."
Get six of our favorite Motherboard stories every day by signing up for our newsletter.