The mining sector has long been at the forefront of autonomous technology, and with good reason.
The operations and equipment take place on a massive scale and the risk to workers is commensurately large. The demand for skilled labour often outstrips the supply, leading machines to step in where humans will not. And resource extraction frequently takes place far from major cities.
In the case of BHP’s Western Australian iron ore (WAIO) business, centred on five sites in the Pilbara region, mining work takes place more than 1000 km from Perth and close to 300 km south of Port Hedland.
“We definitely embrace cutting-edge technology when it’s sustainable,” said Neil Hewson, Principal of Mining Systems Electrotechnology at WAIO Engineering.
“Primarily we use a transition to autonomy – not just in mining, but across all sectors of our business – as a method for harm reduction. Where we’re managing a range of material risks, and specifically autonomous haulage around collision events, the primary driver is to remove personnel from the line of fire.”
The sector was at the forefront of autonomous haulage 10 years ago, for instance. But WAIO’s operations encompass more than that: Hewson oversees mobile mining equipment from autonomous drilling to ex-pit products, which are supplied off-the-shelf by original equipment manufacturers, including Caterpillar, Komatsu, and EpiRock.
“We operate the Caterpillar Command for haulage product and the Komatsu FrontRunner product,” he said. “We operate, per mine, somewhere between 50 and 70 haul trucks across the mine, and then separately we operate a large blast hole drill operation autonomously.”
The equipment is operated by a combination of onsite maintenance and operational teams as well as a remote operation.
“A lot of our mining systems are remote,” Hewson said. “As a team, we’re based out of Perth, but the system, more broadly, is a combination at the moment. We’d like to employ more remotely interrogated data sets, but not all information from these systems is available.”
Ensuring the smooth operation of a sprawling autonomous system such as the WAIO operations requires bringing together a suite of sensors, decision processes and communication protocols. To help make sense of how the pieces of this network come together, create asked Hewson to discuss some of the engineering highlights.
Sensors working overtime
While the early days of autonomy saw existing equipment retrofitted with means of measuring steering and braking information, today’s machines are built for purpose. But it’s the sensor suite that distinguishes them from a comparable large mining truck.
“These products use a GNSS [global navigation satellite system]: the same sorts of tech that you find in your mobile phone for positioning is available to a 400 t haul truck,” Hewson said. “We complement that system with a ground-based reference station so that we can achieve the accuracies that are required to operate machines in close proximity to each other.
“They also use a suite of perception sensors around the machine – a combination of radar and LiDAR hardware devices that detect and range objects that they want to avoid colliding with.”
Coming to a decision
Each of WAIO’s autonomous machines has a suite of engine control modules (ECMs) that enhance the vehicle’s suspension, steering and braking systems, which monitor operations to ensure that control of the machine is predictable. These process the data collected by the sensors and load manage the amount of traffic that travels between a client and a server or application.
“A lot of it is exception-based,” Hewson said. “If there is a reason to communicate that information to other parties, whether it’s other equipment in the vicinity or an application that requires decision-making or an exception that requires human intervention, then that information is passed from client to server and then presented through a client interface.”
If the system detected an unexpected object, for instance, it would result in speed management – stopping or slowing the machine down – and potentially escalating the problem so that a human can intervene.
“I’ve detected something in front of me; I’m not sure what it is, but it appears to be impossible,” Hewson said, explaining the ECM’s decision process. “Can you attend the area and clear the area for me or confirm that I’d like to proceed?”
Unexpected interference
WAIO’s autonomous equipment is managed by a single application and ensuring uninterrupted service is one challenge preventing a shift to fully autonomous operations.
“We need to make sure that our application and all of the dependent services that support it are running in order for any activity to occur on a mine site,” Hewson said. “It’s quite a large investment in making sure that an application is robust, in that it has power and it has supported network connectivity, and it retains services like GNSS, which are outside of our control.”
Even solar weather can interrupt that global navigation satellite system, and the sun’s 11-year solar cycle has recently resulted in bursts of magnetism that have interrupted WAIO’s system stability.
“We look to implement interim containment actions and permanent corrective actions that allow those services to approach that 100 per cent availability metric,” Hewson said.
“Something as simple as a solar flare that travels through space and affects the total electron count of our atmosphere can stop a fleet of 60 trucks in its tracks. There are a lot of dependent services that conventional mining doesn’t have a dependency on that an autonomous system does.”
Border control
Autonomous equipment is designed to operate on its own, but that doesn’t mean it operates unsupervised.
“The basis of an autonomous piece of equipment is that you allow a piece of equipment a permission to do a very small piece of work,” Hewson said. “An autonomous truck is not given permission to think on its own and it’s not truly autonomous at the moment. It responds to a sequence of very short-duration permissions to access a road segment.”
WAIO loads mine models into a GeoServer and then overlays these with a road network.
“That road network is broken into segments of lanes or roads, and each truck is allowed permission for a short section of that road, and as they near the end of that permission, they request permission for the next road segment,” Hewson said.
“Only one piece of equipment can access that road segment at a time, so the assignment engine manages access to those road segments based on where all the other pieces of equipment are.”
This webinar explores how AI and robotics can, in conjunction with human workers, help to reduce risks and create safer environments in heavy industry.
