By Professor Ian Manchester, Director, Australian Centre for Robotics; Director, Australian Robotic Inspection and Asset Management Hub. School of Aerospace, Mechanical and Mechatronic Engineering, University of Sydney, and Dr Andrew Hill, Director of the Rio Tinto Centre for Mine Automation, Australian Centre for Robotics, University of Sydney
It is hard to believe that less than 20 years ago there was very little automation on even the world’s largest, most complex mines. The resources sector of today is a pioneer in the development and adoption of robotics, automation, and artificial intelligence systems, and while much has been achieved there is still much more to be done.
The transformation of this sector since the mid-2000s, to its position today as a global standard for industrial field robotics and automation, is truly remarkable.
Perhaps less well-known is that Australian innovation, scientific breakthroughs, and industry research collaborations with Australian universities have been a significant driver of this transformation. These achievements are proof that Australia does not need to be a passenger in global innovation but can instead be a leader in the development and export of robotic and automation technologies.
Industry-university partnerships
In the early 2000s, academics at the Australian Centre for Robotics (ACFR) at the University of Sydney, made groundbreaking contributions to a technology known as Simultaneous Localisation and Mapping (SLAM). SLAM gives a robot the ability to reconstruct the physical world around it in real-time and orient itself within that reconstruction – much like humans do every day, navigating and operating within their surrounding environment.
Until this point, robotic systems in industry had been largely confined to structured environments, like factory floors, where the scenario and tasks are controlled and predictable. The development of SLAM meant robotic systems could now interact with the unstructured ‘real-world’ and led to an explosion of Australian innovation in industry, with the most notable being in the resources sector.
In the mid-2000s, ACFR academics published a series of papers articulating the vision of an autonomous remotely operated mine. These seminal works and engineering contributions left a lasting legacy and helped propel the Australian mining industry to the forefront in terms of technology investment and adoption.
These innovations also led to an enduring industry research partnership – the Rio Tinto Centre for Mine Automation (RTCMA) between Rio Tinto and the ACFR, which formed in 2007 and is still around 15 years later.
Through this partnership, ACFR researchers and Rio Tinto Engineers have pioneered numerous industry-leading innovations. In the robotics and automation of physical assets, the group has developed autonomous drills, light vehicles, surveying platforms and geological data collection tools. At the fleet level, autonomous train scheduling, truck dispatch systems, haul-road network mapping and material movement planning across a complex mine network.
The proprietary Mine Automation System, or MAS, fuses data from disparate vendors and internally-developed systems to provide a centralised, real-time source of truth, allowing observations of the activities in the mine and providing analysis, insights and support for intelligent decision-making from across the globe.
Data fusion has also been employed in orebody modelling, using probabilistic models to combine complementary datasets as they become available, providing more accurate representations of ore bodies aiding in the planning and design of mine pits.
These innovations only scratch the surface of what is possible.
Striving for completely autonomous operation
It is worth noting for the purpose of this article, that while automation in the resources sector is considered a global standard, we are still very much at the beginning of this journey. We are still yet to meet the inflection point where mines are wholly designed for automation, rather than the incremental automation of processes in a value chain previously undertaken by people.
In other words, a way of looking at mine automation for much of the past two decades is that it has centred on automating equipment that was designed for humans. But what does a mine and its equipment look like if it is designed for completely autonomous operation from inception?
The various answers to this question involve radically different operational modes, safety considerations, and more specialised and efficient equipment.
A popular example of this we use at the ACFR is haul trucks. On first impression one might assume that these tremendous vehicles are as large as they are because it is the most effective way to transport large volumes of crude unprocessed material, but this is not the case. They are large primarily to cater to the constraints of a human operator workforce. Namely, this is a highly-skilled but small workforce creating a potential bottleneck or a point of critical failure, should enough drivers be unavailable simultaneously.
To ensure there are no disruptions to throughput, larger trucks are required to cater to the economics of a small workforce. But if these trucks and other related load and haul equipment are no longer operated by humans, do they still need to be this large? The answer is ‘No’. When we think about a fully autonomous haulage fleet the economics actually favour a smaller vehicle as they have standard dimensions and standard parts, are easier for mechanics to maintain and can drive on standard sized roads that require less maintenance.
Opportunities for automation
When we think like this about the future of mining, examples abound.
Major assets – such as those found on mine sites – require constant inspection, maintenance and repair. This work is almost exclusively done by people and for major industry, this means a whole suite of considerations need to be accommodated into asset design.
Important elements need to be accessible not only for inspection, but designed in such a way that they can be safely repaired by an operator. Stairs, railings, safety cages and restricted areas add a layer of complexity to an operation that would not be required if operations were fully autonomous. Furthermore, many operations have such large safety exclusion areas that they must be completely stopped and isolated to just facilitate visual inspections by operators.
For a mine the scale of those found in the Pilbara, this work stoppage can rapidly accrue millions of dollars in lost production time, and inefficiencies up- and downstream.
There is a tremendous opportunity for future mine processing assets and facilities to be designed free of these safety constraints; where operations and robotic maintenance platforms are co-designed for a fully automated, more efficient system.
A new robotics hub
The ACFR is leading the development of this novel technology in partnership with over a dozen industry partners, Queensland University of Technology (QUT) and Australian National University (ANU) through the newly-created Australian Robotics Inspection and Asset Management (ARIAM) Hub. ARIAM was created to address the impending post-war ‘infrastructure cliff’. As major assets approach the end of their life it is becoming increasingly clear the development of novel robotic and intelligent systems will be essential to their management, now and into the future.
This is also true for the resources sector where mines operate on the timescale of decades, requiring constant inspections and maintenance, as discussed above. ARIAM partners like Nexxis, Abyss Solutions, Emesent and partner resource companies are working to develop novel robotic systems for asset management that take humans out of harm’s way, providing more regular and reliable inspection data in the process.
Optimisation in mining
An area ARIAM is focusing on where we see tremendous innovation potential is ‘whole-of-operations’ optimisation and decision-making for large, dynamic industries such as manufacturing, construction, shipbuilding, transport, and the resources sector to name a few.
Mines such as those found in the Pilbara, and their downstream supply chains, are of a size and complexity so vast it is simply not possible for human based decision-making across all aspects of an operation to result in efficient production and utilisation of resources.
Mining operations are often segmented or ‘silos’, each with its own priorities. While segments do of course work together to achieve a main objective, without a comprehensive real-time evaluation of an entire operation and the state of each segment within the value chain, it is impossible to efficiently determine operational priorities on a granular level at any given moment. For example, the goal of a load-haul operation in one area is to get ore out of the pit as quickly as possible. But if the grade is lower than expected, this will need to be blended with other material. The ‘local’ solution will be to move an excavator to adjust the blend, at the cost of productivity in this shift. But it may be more efficient to stockpile in another load-haul operation with a higher grade, elsewhere in the network of mines, or at the point of sale via penalties/discounts.
It is easy to see how inefficiencies can cascade downstream, compounding opportunity costs at every step. The globally optimal solution can only be determined if the entire value chain is well tracked and modelled, and sophisticated decision-making tools are developed to analyse the situation and potential range of actions. Robotics are essential to achieving this vision as a means of collecting the continuous, high quality data these systems require.
Pushing to meet net zero
Optimisation will become increasingly important as we move into a decarbonised resources sector.
Rio Tinto, BHP, Fortescue and many other major resource companies have already committed to net zero by 2050. The significance of these commitments is perhaps underestimated, not only because the resources sector is responsible for such a large proportion of global emissions, but because the scale and complexity of such a transition is typically underappreciated.
Optimisation will be important; firstly because it will reduce energy consumption by minimising waste, but even more importantly because optimisation will support the design and operation decisions of future mines powered by a completely different energy matrix.
Current design and operations are based entirely on the availability of rapidly deployable energy produced by fossil fuels, but the assumption of constant nameplate power availability will not always hold with renewable energy supply.
So what does a mine powered by renewable energy look like and how does it operate? Can operations be scaled seasonally, daily, or even on the scale of hours to milliseconds, to minimise energy storage costs, or will the generation and transmission infrastructure dictate the schedule? How will the scale of energy infrastructure required be constructed in only a few years?
Optimisation is perfectly placed to help industry make these decisions when systems are complex and ambiguous. Additionally, automation will be a huge competitive advantage for the design, construction, maintenance, and material recycling of a renewable energy matrix. For this reason, the transition of the resources sector to net zero is possibly the greatest opportunity for robotics, automation, and AI innovation.
Many of the opportunities mentioned so far will require additional steps as part of longer-term planning commitments and thinking, and these additional steps are no less exciting or important.
Prioritising data collection
One such step that is central to the ARIAM mission is emphasising the value and need for broader, more comprehensive data collection and sharing when dealing with complex assets and operations. This is especially relevant to mine operations and essential to developing the mines of the future. We cannot optimise what we cannot model, and modelling depends on high-quality data.
Resource companies are amongst the biggest companies in the country and for some, the world. But when we compare data collection and utilisation practises with the world’s largest tech companies like Amazon, Meta, Apple, Google, Microsoft and others, the disparity becomes clear. Although an extreme example between vastly different businesses, the analogy is valuable just the same.
These companies are famous for the amount of data they collect, how they value it, and use it strategically. The general principle is to collect as much data as possible, knowing that its value will become apparent as new analytics and products are developed. It is this mindset that has allowed Apple to develop the most sophisticated and efficient ‘just-in-time’ supply chains in the world, where components from numerous manufacturers, or the subsequent iPhone or laptop, are stationary for no more than a few hours at any stage in the supply chain to the moment it lands in a customer’s hand.
When we think of resources companies through this lens, they aren’t all that different – large, complex operations, with constantly competing priorities and objectives on a global scale. It is especially true given the timescales of mine operations, measured in decades. There is no reason the resources sector cannot operate in the same league as these tech giants, but in order to achieve this, the sector needs to embrace and value data collection the same way the technology sector has. At ARIAM we strongly believe robotics will be central to achieving this as they are the mediators between digital and physical worlds, simultaneously collecting data to construct digital twins of their environment, and using the insights derived from this data to take action in the real world.
It is an exciting time for robotics, automation and artificial intelligence in the resources sector. The sector has already been successful with its adoption of this technology, to the point that one would be forgiven for thinking it is ‘mission accomplished’.
In reality, what has been achieved is a solid foundation from which the mines of the future and all the novel technology that comes with it can be developed and that is something industry can be truly excited about.
The Australian Centre for Robotics (formerly known as the Australian Centre for Field Robotics) is Australia’s largest robotics group and is a global leader in robotics research. Our research tackles deep scientific problems in core robotics technology such as sensing and perception, mapping and insights, planning, control, modelling and optimisation, learning, complex dynamic systems, as well as their applications in complex cyber-physical environments. The ACFR has a long and distinguished history of delivering first in world robotics capabilities to major industry including in resources, agriculture, aerospace and freight logistics sectors. The ACFR is also the lead organisation for the new Australian Robotic Inspection and Asset Management Hub (ARIAM Hub).
