Japanese multinational Komatsu is approaching its 100th birthday, and while the company has only been in Australia for around 50 years, it needed to change the way it operated to ensure it was ready for the next 50 years.
In Australia, Komatsu is mostly interested in mining and construction.
Speaking with ZDNet, Komatsu Australia general manager of construction solutions Todd Connolly said the organisation has innovation to thank for its local success, in 2010 being the first to market with hybrid earthmoving equipment, as well as with autonomous, driverless trucks which he said have been in operation for many decades, and electric mining haul trucks, as some examples.
Komatsu also has a heavy footprint in advanced technologies, with around 16,000 of its 30,000-odd machines in Australia -- ranging from smaller excavators through to mammoth earthmoving ones -- boast Internet of Things (IoT)-style sensors that stream data mainly via satellite, but also through GPRS and other means.
Komatsu also operates three oil laboratories in Australia which Connolly said provide very deep analysis of machine health and another dimension to the data the company uses to help maintain machines.
"We're quite data rich in that sense," Connolly
said. "Our objective is really to turn data into actionable information."
According to Connolly, this is what led Komatsu down the path of looking at unifying and aggregating that data.
"And more importantly, turn it into information that our customers can use to get a better return on their investment from their Komatsu machines and our own people to use to drive efficiencies and better business processes within our business," he added.
Komatsu originally went down the path of unifying a series of data sources, driven by having availability and visibility of its data.
The company has three major sources of data: Its ERP system, equipment telemetry systems, and laboratory data.
"The strategy was to unify all of that data into a single platform," Komatsu Australia general manager of business technology and systems John Steele told ZDNet.
"When we initiated that, we sort of did that in an on-premise context and then it became how do we actually manage all of that data and the amount of compute capacity that it was requiring."
Komatsu as a result last year turned to Microsoft, deploying Azure SQL Database Managed Instance, combined with Discovery Hub and Power BI, to help process its vast amount of data.
See also: Microsoft Azure: A cheat sheet (TechRepublic)
"That made sense to us for a number of reasons, partly the amount of compute capacity that we can put online compared to what is available in our on-premise environment, and also the ability to grow that environment rapidly as our data grows ... and then also the ability to leverage the Azure platform and the capability within Azure with respect to elements like AI," Steele continued.
Where artificial intelligence (AI) is concerned, Komatsu Australia analytics architect of business technology and systems Nipun Sharma explained the first instance of Komatsu enabling AI would be for data collation.
"We are collating -- that's the unifying part of it, bringing all the machine business systems and condition reports -- a lot of this information that is disjointed, bringing it together," Sharma said. "Having that platform in Microsoft Azure, we plan to leverage a lot of functionality available in Azure itself in terms of AI and machine learning. So we are doing some prototypes, looking at Azure [Managed Instance] cloud offering again, we are also trying to leverage a bit of process automation ... so that's all happening right now, using libraries and other things available on the Azure platform."
The predictive maintenance proposition
With all of the information at the machine level, Connolly said the organisation wanted a way to pre-diagnose machine condition before sending a technician to provide a repair.
"The challenge we had historically with doing that pre-diagnosis was that the data being spread across so many platforms and systems meant that preparing for that could take well over 40 minutes," he said.
"So initially what we did was we built some databases all in SQL and used the SQL stack to come up with a faster way of compiling that data. That was a part of a process that we broadly called Fix It First Time. So when we go to a machine we can fix it as quickly as possible.
"But of course, once we started on that journey it meant that we needed to find more scalable ways of doing that and we wanted to go from just collecting information to actually making suggestions to our technicians about how they should go about that repair. I think that's probably one of the bigger opportunities that we have going forward with the tool sets that we're working with now is to use the data to suggest what the repair mechanism should be and what tools should be taken."
In addition to helping customers minimise downtime of their equipment through predicting maintenance and repairs, Connolly said Komatsu can also help customers get the most out of their investment.
"For instance, a dump truck exists to move earth from point A to point B, but there are many variables which will dictate the efficiency of that. And one of the larger variables is how that truck is loaded. So by using data that allows us to understand the loading accuracy and the cycle time and aggregate that information across the entire fleet of machines," he explained.
"We can, through creating scoreboards and other mechanisms like that, help our customers identify where they should focus. So it might mean needing to focus on training of the loader operator, or pointing out some design changes that may be needed for the road or the quarry in which they're working. Or indeed it might mean changes to the machines themselves."
"This entire process is going to make our business processes more intelligent and that's what we're looking for into the future," Sharma added.