The Journey to Autonomy - Part 1
Read time: ~3min
In this series we will explore some potential paths from present day to autonomous operations, and the challenges and requirements of achieving autonomy(x).
Why a journey…
The journey to industrial autonomy goes beyond automating a haul truck or changing a process control system from PID to MPC to AI. Autonomous operations are idealised for many reasons, principally because we know there are lots of mistakes that are preventable, and we know we need to change as society and our incoming workforce changes, and we want to increase performance while constraints on how we operate are ever increasing, and we want to take advantage of technology and capabilities that weren’t available even 2 years ago… The rationale for many organisations are multiple and often a different stakeholder will emphasise a different reason. Regardless of the true or root causes or rationale for or against, some organisations will pursue autonomy at great cost, while others will wait too long and be disrupted. The next Kodak will very likely be disrupted by autonomy, not just “AI”.
What is autonomy?
Autonomous operations need not only mean unstaffed, however, and we propose a useful proximate objective as being autonomy(organisation) = x, where x = the years where a majority of decisions in that organisation are still made by humans.
Why decisions as the measure of achievement? Because while human decisions are still in the majority, the paradigm of autonomy will still be viewed as assisting humans. When the coin turns over and our decisions are in the minority, we’ll then realise that autonomy is about humans assisting machines to achieve things we could never achieve alone, in the same way that we couldn’t move faster than a running horse, or fly, or melt iron, or nay of the others things we can’t do without machines. Autonomy can therefore be defined as the horizon at which human decisions pass from primacy to assistance, in the same way that our legs passed from motive power to pressing the pedal down harder.
Why organisation as the fundamental unit of measure? Because in an increasingly tech enabled world, some organisations in a given industry may achieve autonomy next year, and some will never achieve it, and inside organisation some pockets will achieve autonomy next year, and some pockets will never get there. As per above, however, when a majority of decisions in an organisation are machine dominated, the entire culture of the organisation will be irrevocably changed. The organisation itself may only ultimately achieve 51%, or 60%, or 70% autonomy, but it will be an autonomous organisation, viewing itself as such, and we believe that is a major and almost certainly irreversible milestone to cross.
What is autonomy, really?
When we think of autonomy in mining, we immediately jump to autonomous vehicles. The advances in autonomous haul trucks in the last 3 years have rightly captured the interest of many, although there is some conflicting information on benefits vs remote operation vs traditional. As a new technology, we should expect ups and downs that will constitute the learnings we need to adopt and mature this new technology, and the ways of working that need to go with it.
After haul trucks, the next most common autonomous theme is processing plants, in particular control systems. In both cases, and for perhaps 90% of social media posts, the specific challenge being addressed is autonomous equipment control. Beyond equipment control, however, there are other types of autonomy, and there are degrees of autonomy that will impact mining, as much or more so than direct control.
It’s all a journey…
The awful use of “journey” made us very hesitant to even broach this subject, but some things are a journey; we can’t, or shouldn’t, leap right to the final destination, and along the journey, there are challenges for the protagonist to overcome, lessons to be learned and victories to be celebrated. Let’s dive into the journey and see what we learn.
The journey to autonomy
The journey to autonomy(x) for any one organisation will encompass many gradations, but we can hypothesise some likely milestones:
Assistance
Simultaneous autonomous operations
Zero entry
Unstaffed
The first and most obvious, and perhaps most unpalatable to many, is the autonomous, un-staffed mine. Un-staffed = no people. Machines run the mine through prescriptive decisions based on predictions and analytics, and non-machine decisions are made by remote humans. Of course, there are serious issues with maintaining equipment and robots with robots, and many things that benefit from the dexterity and coordination of the hand/eye. Similarly, 99.9% of the physical world is designed & built for “human scale”, so in order to enact an unstaffed operation, humanoid robots will need to proliferate and existing equipment will need to be heavily modified. And come shutdown time, it’s very unlikely that many companies will have invested sufficiently (with so little reward for so long) to make a shutdown unstaffed. Similarly, unplanned downtime that requires international robotics experts to fly in and troubleshoot, which could be solved by a 40-year-old fitter who lives locally, is likely to be solved by said fitter. The conversion of an existing operation to an unstaffed mine is, we believe, improbable in the next 20 years in most organisations, though a tier 1 at the top of the cycle will likely attempt it.
Greenfield's unstaffed may be a slightly different story, and there are some non-traditional environments where unstaffed makes a lot of sense: space mining and deep sea mining. In both of these environments, unstaffed has to be the goal, and reducing Full Time Equivalent (FTE) staff to 10% or 5% or 2% of an equivalent tonnes/size operation may be the tipping point between profitably opening up outer and inner space development at scale vs only doing it for show. There are other benefits to autonomy besides cost and reducing physical human errors, but they are beyond the scope of this article. Suffice to say, reaching for un-staffed in deep sea and space may provide catalysts for innovations that greatly benefit terra-firma mining, and enable largely unstaffed future operations to be built.
A step back from unstaffed is zero entry. In this scenario, human FTEs are drastically reduced or eliminated from all hazardous areas, e.g. “underground”. I use quotations because, in a similar view to unstaffed, there’s daily operations, and then there’s maintenance and troubleshooting and shutdowns and…. things going wrong. With the need to hold technical staff for those lost hours, many sites will maintain core staff at the site to troubleshoot and provide sufficient support and coverage for 24/7/365 operations. Zero entry operations are perhaps a 30-50% reduction in FTEs at the site compared to conventional, with extensive use of remote operations centres.
A step back from zero entry is simultaneous autonomous and manual operations. As is being practised now, dividing sites up into Autonomous Operations Zones (AOZ) gives sufficient separation of staff and autonomous machines to operate and learn. Simultaneously, selective non-mobile plant autonomy is being experimented with across analytics, control and optimisation systems. The rise of predictive analytics is driving much of this, and the cross-overs of predictive support to enhance human decision-making, and prescriptive autonomous actions, is playing out in experiments and limited operational aspects across the world.
Underpinning all of this are the experience and rules-based systems upon which we have built the physical world. Even at the ultimate unstaffed site, it is currently inconceivable that there are no engineers, planners, or executives making decisions, undoubtedly supported by smart machines, that direct the courses of events at an unstaffed site. These humans, and the many humans across the supply and value chains, will also benefit from machine-assisted decisions and actions, and will also need to interact with the autonomous mine and support and enable un/lower staffed operations. Of course, AI maximalists will dispute this, saying that we can do away with all humans for many aspects of life. Undoubtedly, they are true, in the same way, that undoubtedly we’ll build a permanent base on Mars – it will happen, and it’s not infeasible, but it’s sufficiently far away as to not materially impact our current ways of thinking, except to provide a signpost labelled “the future”.
Most interestingly, none of the above milestones necessitate autonomy(x). An unstaffed mine with 100s of remote workers and large local contracting teams that regularly descend onto site may end up looking like a very large hand puppet. This outcome is probably to be avoided as it loses much of the benefit of autonomy and reduces the agency and context of the humans who make the majority of decisions.
Conversely, all of the above milestones could conceivably follow autonomy(x). An organisation that disrupts itself sufficiently and incrementally could still have manual operating zones (e.g. shutdowns as a temporal rather than spatial zone) while the majority of operations are autonomous.
Pragmatically, autonomy(x) is likely to be necessary to achieve sustainable zero entry, and occur after substantial process plant automation, and perhaps at a hub type operation that is close to other sites and with existing close-but-off-site logistics and support. Much will depend on the real cost benefits of autonomy(x) across a sufficiently long time line. Reducing AISC for a few months only to lose all that benefit in 2 extended unplanned downtimes is every miners fear when embracing innovation.
How long will it take?
It’s a question on many people’s minds, but it’s perhaps not the right question to ask. The journey to autonomy(x) is a journey of deliberate change. Deliberate change is the result of ideating, executing and managing many new and unknown events. If we think in terms of events and change, rather than linear time, then the answer is: as many events as it takes to move the business from being experience led to prescription driven. Ultimately, the adoption of prescriptions (machine recommended actions), is the final arbiter of all autonomy progress. To adopt prescriptions at any meaningful scale (remember we’re anchoring on “majority”), the journey an organisation takes is to let go of existing experience led processes, structures, delegations of authority etc and steadily replace them with more repeatable, data informed actions.
X will be driven by the speed at which the organisation can let go of the existing and replace it with the new.
In part 2…
In part 2, we’ll take a closer look at this journey from experience led to prescription driven.