Peter Corke is well known for his research in vision-based robot control, field robotics and wireless sensor networks. He is currently Professor of Robotics & Control at Queensland University of Technology.
Corke is editor-in-chief of IEEE Robotics & Automation magazine and a recipient of the Qantas/Rolls-Royce and Australian Engineering Excellence awards.
Prior to QUT, Corke was senior principal research scientist at CSIRO where he founded the Autonomous Systems laboratory, a 50-person team undertaking research in mining, ground, aerial and underwater robotics, as well as sensor networks.
He subsequently led a major cross-organisational "capability platform" in wireless sensor networks. Corke was the keynote speaker at the recent National Instruments Autonomous Robotics Competition in Sydney and took time off to speak with PACE magazine.
What interests you about robotics and mining?
We’ve got to produce an increasing amount of ore product and just don’t have the workforce. The only solution is to make our existing workforce more productive and automation is the tool to accomplish that.
A technology possible right now is self-driving trucks used in open pit mines. We developed the technology with CSIRO and licensed it to Caterpillar. It’s now in production at a lot of mines around the world.
There’s also been work done on the automation of excavation on surface mines covering draglines, rope shovels and hydraulic shovels. Engineers have been looking at how automated shovels can automatically dump material into trucks. Automation of blasting is another thing I looked at when I was at CSIRO. How do you automatically load explosives into predrilled holes?
What are some of the challenges engineers are working on right now?
I think the biggest challenge is building systems that are reliable. We need to involve companies that are in the business of delivering automation solutions into an industry like mining.
There are lots of good ideas in research labs and universities, but getting it out into the field is going to require companies well skilled in automation technology and who have a good support network in some very remote parts of the country.
There’s no point replacing a manned system with an autonomous system which then breaks down for want of spare parts.
Do you see any leading edge technologies that could be applicable to Australia’s mining sector?
A lot of the good mining automation research is actually happening seems to be happening in Australia – at CSIRO at the University of Sydney and a couple of universities in the US and maybe Europe. There’s not many people working in this field.
There has been a natural conservatism within the mining industry that they don’t want to go this way. But in Australia right now, because of this crazy resource boom that we’re having, they’re going to have to embrace automation in a big way.
A lot of innovation is driven by mining companies rather than mining equipment manufacturers. The Rio Tintos or BHP Billitons want the technology, but their vendors oftentimes seem a little reluctant to provide the autonomous systems these companies are increasingly seeking.
Maybe they don’t have the skills in-house. Caterpillar would be an exception as they have a huge automation program and now own a large number of mining equipment vendors.
[Peter Corke has just published a book: Robotics, Vision & Control which is a walk through the fundamentals of mobile robots, navigation, localisation, arm-robot kinematics, dynamics and joint level control, then camera models, image processing, feature extraction and multi-view geometry, and finally bringing it all together with an extensive discussion of visual servo systems.]
Have you developed any non-mining systems as yet?
I’ve just started at university, at the beginning of 2010, so we have one project up which looks at automation in broad agriculture.
So that’s wheat and chick pea crops and particularly chemical weeding. The current approach is to build bigger and bigger machines with huge booms, spray lots of chemical over the landscape.
Maybe we can do it smarter – build many more smaller machines that act more locally, maybe not waste as much herbicide.
This is also a more robust solution. If you’ve got one big machine and it breaks you’re in trouble. If you’ve got 10 machines and one breaks, you’ve still got 90 per cent capacity.
What would you consider to be your major contribution to mining automation?
I’m really proud of two things. One is the underground navigation system that we patented and licensed to Caterpillar.
This allows an unmanned vehicle to drive at pretty much its maximum speed safely through underground tunnels and get quickly from point A to point B and back again. That was a really nice piece of work.
The other one involved automating a dragline excavator. We were able to demonstrate that this massive machine could be automated to dig as well as a machine controlled by a human operator.
When the mine guys looked at the machine doing its work they couldn’t tell whether the computer was controlling the dragline excavator or whether one of their operators was driving it. That was pretty impressive considering they were very sceptical when we started.
We did an awful lot of field work. I think over the course of one two experiments, we moved 250,000 tons of dirt. I think we can say it’s the world’s largest exercise in robotic excavation or dirt moving. That record will stand for a while.
What unique technologies did you use?
The innovation was mostly it was in the sensing side. How do we understand the state of the environment? Where is the dirt?
Where should we place the bucket onto the ground? When the machine is moving, where is the bucket with respect to the dragline because the bucket is suspended on cables?
We need to understand where that bucket is, because it can swing from side to side with respect to the dragline.
There are also elements of the operator’s skills that we needed to capture. Although the machine looks very slow and cumbersome, there are some aspects of its operations that are very important.
The operators make it look very easy and there was a lot of conversations with operators, a lot of observing them to work out what were the elements of this skill and then capturing that in a large computer program that would emulate what they did.
Then we’d run that; they’d look and critique it, saying things like ‘no, you did that too late’. So then we went back, modified the code and tested it again.
It was a nice interaction between a bunch of technologists standing behind the operator while he’s doing his work and then the operator watching over us when our computer is doing its work.
We overcame their scepticism and it’s nice, but I’d love to see it in production. So that means we need to find a company that wants to license the technology, build it, install it in draglines.
There was interest at the time, but then the GFC hit. Fortunately, interest in mining productivity is ramping up again so its time will come soon.