As a result of the mining industry's widespread push to reduce costs and increase productivity as safely, efficiently and profitably as possible, autonomous loading and hauling has formed one of the principal R&D thrusts in underground mining in recent years.
Moving to fully automated vehicles forms the next logical step from the already well-known tele-remote-control systems for load-haul-dump (LHD) vehicles and offers some significant operational advances over this established technology.
While teleoperation achieves safer working conditions, it has its limitations; remote operators cannot drive the vehicles as quickly as an on-board driver and, denied the sound and 'feel' of the vehicle to alert them to changing conditions, they are forced to rely heavily on video cues.
With the 'haul' component being the longest element of the LHD work-cycle, there are obvious advantages to automating this aspect of the operation. It is an attractive proposition, not least because it promises greater safety, higher productivity and lower maintenance costs – ensuring that the automation issue remains high on the development agenda.
This move towards automated underground navigation gained further momentum in July 2007, when Atlas Copco Drills awarded a contract to MacDonald, Dettwiler and Associates (MDA) to provide navigational systems for their LHD vehicles.
The presence of MDA in this arena should come as little surprise; MD Robotics – part of the same corporation – has an enviable track record of developing remote control equipment for challenging environments, having designed the robotic arms for NASA's space shuttles.
The company's advanced guidance and control technology senses tunnel walls, calculating the LHD's position against system-generated maps and enabling the vehicles to travel through the mine without the need for human input.
Autonomous underground navigation
The idea is not a new one. A variety of approaches to autonomous underground navigation have been used with varying degrees of success around the world over the years – though all have relied on the installation of some form of additional infrastructure to guide the vehicle.
Systems based on external physical guidance – such as light ropes, inductive wires or reflective tape – have a number of serious drawbacks, principally the difficulty and expense of guide installation, maintenance costs and inherent inflexibility.
Perhaps their most important limitation however, is the fact that none permit high-speed operation.
With a driver on board, speeds of 20km/h to 30km/h can be achieved; the older generation of navigational systems seldom provide more than a fraction of this, since they all have limited ability to look ahead, only being able to respond to the guidelines in their immediate vicinity.
In addition, none of the older systems offer obstacle detection – an obvious weakness in often changing and unpredictable underground conditions. The latest wave of solutions relies on robust automation systems and innovative software applications to provide full functionality while removing the driver from the cab.
Underground mines – despite their often labyrinthine nature – are relatively simple environments for robotic navigational systems. Although the topography of the tunnels and intersections provides a wealth of information to process, this is as nothing compared with the computational challenge faced by fledgling autonomous automotive technologies intended to navigate through heavy traffic, for example.
Mining applications face no major technological barriers and according to Australian research lab the CSIRO, within ten years, advances in integrated, real-time processing will open the way to automated mining operations and autonomous equipment, especially in more hazardous environments.
However, the advantages are not simply limited to health and safety concerns around potentially dangerous sections of underground workings.
Planning and scheduling can benefit, while having no drivers removes the down-time of shift changes, opening up the possibility of true 24/7 operation. Crucially, it also promises to bring directly improved efficiency and greater productivity.
During system trials it became clear that autonomous trucks can be run at speed with less risk of striking the tunnel walls than human-driven counterparts and though their individual cycle times may be slower, output production actually improves.
Part of the reason for this is the consistent uniformity of each LHD run, which coupled with computer optimisation of time use ensures maximum returns.
In the same vein, bringing vehicle operation under system control means that the driving is always perfectly matched to conditions, gear changes timed exactly and excess revving avoided – reducing fuel consumption and significantly lessening maintenance requirement. Unsurprisingly, these paragons of mechanical virtue have not always found universal favour amongst mine workers themselves.
The commercial challenge
To be commercially viable, any system to provide LHD vehicles with autonomous navigation inevitably needs to fulfil a number of criteria, including ease of set-up, low maintenance and a minimal demand for additional infrastructure.
In addition, it should not require extensive mine mapping – instead using maps created by the system itself once installed – and it needs to allow vehicles to self-guide without human intervention at comparable speeds to those reached with an on-board driver.
Reliability and safety requirements are clearly also high, making it essential that the system has robust self-diagnostic fault detection and fail-safe mechanisms in place.
With such a potentially lucrative commercial opportunity, Atlas and MDA are not the only players in this particular game. Other companies have also been busy in the sector, particularly Caterpillar Elphinstone and Sandvik Tamrock producing the MineGem and AutoMine systems, respectively.
MineGem was developed by Dynamic Automation Systems (DAS) – a joint venture between Caterpillar Elphinstone and Lateral Dynamics, building on the results of work originally begun in 1996 by a team from CSIRO and the University of Sydney. However, once it came to commercialising the technology, it was AutoMine that was to claim the success and when the largest underground mine in the world – Codelco's El Teniente – installs your system, it has to say something.
A number of mining firms, including Inco, Kafi und Salz, LKAB, Noranda and WMC had run trials with automated LHDs, but in June 2004, the El Teniente Mine in Chile became the first mine to use an advanced autonomous system for large-scale production.
Since then, Sandvik has notched up further successes with its AutoMine systems at the Inmet Mining's Pyhäsalmi Mine in Finland (January 2005), De Beers' Finsch Mine in South Africa (August 2005) and most recently, in June 2007 at the Williams Mine in Canada.
The AutoMine System
The AutoMine system relies on lasers mounted on the vehicles to scan ahead, detecting the tunnel profile and enabling the LHD to establish its position in milliseconds – which avoids any need to mark the route with reflective strips or RFID tags.
The topographical information obtained allows the system's tunnel map to be continuously updated – requiring a large volume two-way data flow between the autonomous vehicle and the system over dedicated wireless local area network (WLAN) radio communication.
Despite the sophistication of the approach, conventional and easily available computer hardware is used to provide system control and many of the components are standard units specifically chosen for their proven durability in similar applications.
The mining industry has a long history of embracing automation and the march of this particular technology – which has already brought autonomous navigation through its early years of development to become a productive, real-world tool – seems set to continue. Whether it will eventually become commonplace, however, remains to be seen.
Despite their many advantages, these systems do not come cheap and mine operators will inevitably have to be convinced that the financial returns they promise can be delivered.
Moreover, there is also the wider question of individual mine suitability. Automated systems require separate sections of the underground workings to run in, where human access can be strictly controlled and relatively few large mines have the type of block caving ideal for these high-spec, fully autonomous technologies – and can justify their cost.
This does leave one area in particular open to possible further development in the future – a lower-cost and less-sophisticated system which maintains many of the key operating benefits, but dovetails more easily into situations which currently use conventional remote control.
Offering the opportunity to automate single LHD vehicles without the expense of full autonomous technology – and retaining the option of manual override to revert the vehicle to remote control mode as and when necessary – allows the needs of smaller mines to be met.
According to a study commissioned by Caterpillar, changing from conventional teleoperation to such a single vehicle automation system should achieve payback in around three years and could provide more than a 35% increase in productivity.
While safety and efficiency are major drivers on the trend towards automation, in the end it is inevitable that the future of autonomous mine navigation will stand or fall on its commercial performance.
If the figures do work out as well as they promise and the costs of making a mine safe and suitable for humans to work can be significantly reduced, the future of the technology should be assured.