Three partners have created a robot
capable of recognising and selectively cutting minerals in a potash
mine. The robotic mining machine is a conventional roadheader controlled
by a computer. A video image of the rock face is processed to map
the distribution of minerals which are then cut in a sequence planned
to minimise the final cost of bringing the potash ore waste to the
surface. The robotic cutter proved more efficient than human operators
in discriminating between ore and waste, with important economic and
The work of the miner is one of the most
difficult and dangerous in the world. It calls out for automation,
yet robotic technology has made little headway in the mining industry.
Until now there has been no substitute for the hand and eye of a
human being in the efficient extraction of minerals.
It is not hard to see why. Most industrial robots, such as those
on car assembly lines, work on repetitive tasks in carefully structured
and controlled environments. Conditions in a mine, on the other
hand, could hardly be more different - heat, jolts, vibration, dust
- with constantly changing surroundings as the cutting proceeds.
Moreover, the miner - whether robot or human - must recognise the
difference between seams of valuable ore and worthless waste, and
ensure that one is extracted and the other discarded.
Despite these challenging problems, this project, involving three
partners from Spain and France, is a continuation of a project funded
by the EC Raw Materials Programme to develop a robotic cutting machine
for use in mines. The project was centred on the Potasas del Llobregat
potash mine in Sallent, Spain, which supplied an existing 'roadheader'
mining machine to be modified for robotic working. Most of the development
work was done by the project leader, AITEMIN, a mining technology
research centre in Madrid, in cooperation with LAAS-CNRS, the national
Laboratory for Automation and Systems Analysis at Toulouse.
The roadheader was a 95-tonne Vöest-Alpine AM-100, a tracked
vehicle with a cutting head attached to a long boom. In manual operation
the vehicle is positioned in front of the face and the operator
cuts the rock by eye, moving the cutting head with controls in the
cabin. About 40 cubic metres of rock can be cut before the vehicle
needs to be moved forward.
For robotic control an industrial VME computer was fitted on board
the vehicle, protected against dust and cooled against the 38°C
temperatures found in the mine. It was linked for monitoring purposes
to another VME on the surface four kilometres away.
Video cameras create a face map
An early problem was how the machine could 'see' the rock face
and identify the different minerals. After experiments with mapping
gamma radiation over the face (potassium salts are slightly radioactive),
and investigating the use of microwave absorption and other techniques,
the team decided that visual mapping was likely to be the fastest
and most efficient method. To cover the 45 square-metre face in
sufficient detail (some of the layers of clay between the seams
are less than a centimetre thick) four video cameras were positioned
on the roadheader, grouped in pairs, each pair with its own integral
lighting. A major success of the project is the creation of software
to fuse the four overlapping images to make a single 'face map'
of the rock in front of the machine.
Next, the computer analyses the map to identify the different minerals
and their precise boundaries. The potash ore in the Sallent mine
is a reddish mineral known as sylvinite, a form of potassium chloride.
It is found in two layers separated by orange-coloured halite (rock
salt) which is not needed. A poorer quality potash ore, carnallite,
is also present but regarded as waste.
The computer finds the boundaries between the various seams, and
identifies them primarily by colour and texture (sylvinite is redder
and more coarsely layered than the others) with the assistance of
neural networks. The final face map then shows the precise location
of the various minerals in front of the roadheader.
Based on the map, the planning module decides the most cost-effective
way to cut the minerals so that the final cost per tonne of mined
potash is as low as possible.
Catching up with the robotic miner
In the Sallent mine the cut minerals are carried away from the
face by shuttle trucks and the sylvinite taken to the surface by
conveyor belt while the waste is deposited in worked-out caverns
within the mine. A major problem of efficiency is that the roadheader
can work faster than the trucks can remove the minerals, and it
often has to stop cutting to wait for the next truck. One aim of
the planning module is to minimise the number of trucks needed by
ensuring that, as far as possible, each one is filled to capacity
and the waiting time is reduced.
Finally, the computer controls the movements of the cutting head,
much like a robot arm, excavating the different minerals according
to the plan. Normally the rock salt is removed first, followed by
The robotic roadheader is capable of excavating the entire face
unattended - some 35 tonnes of rock - but it has to be paused manually
to ensure that the trucks are in position. Where the seams are uniform
the machine can automatically move forward for the next cut, but
manual positioning is still required where there are changes of
seam thickness and slope. The problem of automatically navigating
a roadheader along a seam has not been addressed in this project.
A money-saver with environmental benefits
Trials at the Sallent mine were a great success. The robotic roadheader
proved to be better than human beings in discriminating and cutting
the different minerals. The economic advantages of this are clear
- no mining company wants to spend money bringing waste material
to the surface, and clean ore naturally fetches a higher market
price than contaminated ores. The environmental benefits are equally
obvious, as there is ultimately less waste to be disposed of on
While research into automated mining is going on elsewhere in the
world, notably Canada, the partners believe that their robotic roadheader
is the most advanced of its kind.
Possible spin-offs include using the image-processing software by
itself to make geological analyses from photographs of mine faces,
and using positional data from the roadheader to make three-dimensional
models of mine workings with accurate estimates of the amount of
each mineral present.
An idea whose time has not yet come
The project has been stalled since the EC contract was completed
in 1994. With the European mining industry suffering from recession
and restructuring, industrial development of a robotic roadheader
is not seen as a high priority. AITEMIN believes that robotic technology,
while desirable, may be too sophisticated for an industry traditionally
reliant on simple, robust and easily maintained equipment. Ten years
hence the outlook may be different, but for now the robotic miner
remains an idea whose time is still to come.