More than 600 electric rope shovels are hard at work in coal mines around the world.
A reliable automation system would give operators of these workhorses of the mining industry a welcome combination of higher productivity and increased safety.
The big question of course, is how to do it? Electric rope shovels have been around for a long time, but currently, there are no systems available that can demonstrate complete automated cycles including digging, collision-free swing cycles and truck loading.
Over the past several years, CSIRO ICT researchers, Kane Usher and Matthew Dunbabin based at the Queensland Centre for Advanced Technologies (QCAT), have been developing and demonstrating solutions to the rope shovel automation problem.
Funded by the Australian Coal Association Research Program (ACARP), in 2006 their research demonstrated full autonomous digging and multi-pass truck loading using a 1/7th scale electric rope shovel and truck system.
Modelled on a Komatsu 830E truck, the system is equivalent to a 2-3 pass fill using the model shovel.
The automation system consists of a series of modules for digging, collision-free swinging, finding the truck, and selecting and optimal load position.
To allow the autonomous excavation and loading systems to “see” the environment in real-time, two range scanning lasers were installed.
The first scanner monitors the dig face along the dipper arm, allowing for dig planning and execution, and dipper tracking.
The second scanner, offset from the first, allows unobscured scanning of the dig face and surrounding environment.
As the shovel rotates, the laser system generates high-resolution digital terrain maps of the environment, including the truck.
These maps contain sufficient detail for planning optimal dig and loading locations, obstacle avoidance and for estimating the volume of material excavated.
The autonomous digging system scans the dig face and plans a trajectory to fill the dipper.
The system then controls the crowd and hoist drives, and monitors and corrects for dipper stall.
The system also monitors in real-time the dipper “fullness” during the dig cycle and when full, disengages from the bank and commences the swing phase to load the tray.
To determine the swing path dipper trajectory, the system uses the dynamically updated digital terrain map to ensure the dipper and shovel body are free from collisions with the bank, tray and of course from self-collisions.
The truck identification system automatically determines the location and orientation of the truck and the loadable region of the tray in 3D space in real-time as the shovel rotates.
The optimised truck loading strategy then uses the real-time generated 3D map of the truck tray and calculates the optimal placement of the dipper taking into account the current load distribution, spillage, manufacturer’s desired loading profile, and multi-pass fill requirements.
On swinging back towards the bank for the next dig, the system then determines if the tray is full or whether another pass is required before notifying the operator.
The system has been extensively tested on the 1/7th scale system including a demonstration in which 102 autonomous excavation cycles were performed, ranging from shallow cuts to deliberate aggressive penetration of the dig face.
During the experiments, the controller was able to complete every cycle without dipper stalling.
For the purposes of these trials, the translation of the entire machine was not automated.
Therefore the automation system evaluated the available (reachable) material around the shovel and advised the operator when to move the machine. It also suggested a preferred direction for the best future dig zone sequencing.
The performance of the tray identification and multi-pass loading algorithms were also evaluated during this trial.
Here the tray orientation and position with respect to the shovel was varied after each complete tray loading.
Throughout the trials, the system demonstrated reliable truck and tray identification and filling.
A primary measure of system performance is the time taken to complete a pass, that is, from the start of one excavation to the beginning of the next including digging, swinging, tray identification, dumping and return to dig point.
The performance of the automated excavation and loading systems was compared against productivity measurements from two operational machines.
Results show that the system is capable of producing complete pass times that are equivalent and potentially better than current manually operated machine performance.
Although these systems were developed and refined on an electric rope shovel, they are equally applicable to hydraulic shovels.
In a production environment, these systems have the potential to improve both damage control and productivity.
Improvements in damage control will come through reduced collisions, as well as mitigation of stall while digging, while improvements in productivity will come through providing rapid and consistent load pass times.
The project has clearly demonstrated the possibility of reliable and practical excavation and truck loading using autonomous electric mining shovels.
The next step is to trial the technology on a full size machine at a working mine.
Towards this end, the CSIRO and CRCMining are currently working jointly on the ACARP funded Shovel Load Assist Project. This project will use elements of the CSIRO technology presented in this article, combined with technology developed by the CRCMining to demonstrate an operator-assist system for truck loading on a full-scale electric rope shovel.
Kane Usher
QCAT
07 3327 4464
Kane.Usher@csiro.au