Featured Product
This Week in Quality Digest Live
Innovation Features
Edd Gent
Increasingly visible and accessible, but with caveats
Matt Fieldman
German system offers tips for U.S. counterparts
Belinda Jones
Reaping the benefits of manufacturing intelligence
Marni Baker-Stein
Closing the education-to-hiring gap

More Features

Innovation News
VSL hosts special edition of show at new center in Rotterdam
Latest line touts comprehensive coverage, ease of use
Same price, double the resolution and accuracy
Low-cost prevention of catastrophic failures
Diagnoses LVDT failures, disconnects, or short circuits
Using noncontact, interferometric technology
Quality doesn’t have to sacrifice efficiency
WFL continues a legacy of success

More News

Mike Richman


Field Report: HxGN Live

Smart factories run on information, not data

Published: Thursday, June 29, 2017 - 11:03

If there was one key takeaway from Hexagon’s impressive and impressively large user conference, styled “HxGN Live,” which took place earlier this month, it’s that finding actionable information, not merely acquiring mountains of data, is the key to developing a truly smart factory. “It’s always about information, never about data,” Hexagon CEO Ola Rollen confirmed during his opening keynote at the show.

Context is the most important factor necessary to unlock information from data, and the human brain teases out context exquisitely well. Machines, not so much—yet. In areas as diverse as autonomous driving, resource management, building, geospatial positioning, and smart manufacturing, however, Hexagon is engineering a future in which machine learning will establish feedback loops to increase efficiency and reduce response times. Rollen credits human intuition with our unique problem-solving abilities; creating a kind of digital intuition and embedding it into machine-based systems is the endgame. It’s an open question as to how long before that ambitious goal comes to reality.

Joseph M. Juran famously applied to the quality industry the Pareto principle, in which 80 percent of effects are derived from 20 percent of causes. The smart factory takes this conventional wisdom and pushes it to the logical conclusion: That far more than 99 percent of important information can be found in much less than 1 percent of quality data acquired by automated equipment on the factory floor. And it’s within this narrow slice of super-pertinent informational bits that Hexagon’s Manufacturing Intelligence division delivers ultimate value to its industrial customers.

The vision

One of the major announcements coming out of the show involved the model factory that Hexagon is now building in Hongdao, China. Scheduled to open in 2020, this 560,000-sq-ft facility, which will carry a price tag of approximately 90 million euro ($102 million), will put into reality many of the aspirational notions that Hexagon management talked about at this year’s show.

From Hexagon’s perspective, there are several good reasons to launch a factory of this type. First, it will function as the company’s own manufacturing facility, where Hexagon Manufacturing Intelligence will research, develop, build, and test its next-generation equipment for industry. The factory will also allow the company’s engineers to explore better ways to integrate their various hardware and software sensing and feedback tools to achieve Rollen’s dream of a kind of digital intuition within the smart factory. Most important of all, it will serve as a model facility and showcase, where Hexagon customers can come and see for themselves how the digital thread of information can transform their own processes.

After all, Hexagon is first and foremost a customer-centric organization, and the mission to serve customers is never lost on top management. The Hongdao facility will allow the kind of immersive, show-and-tell experience that will really help the company’s manufacturing customer base understand the possibilities of the smart factory in the real world.

“When I talk with customers, at this event or throughout the year, they always want to know, ‘How can you help us to make sure that our processes are more seamless?’” says Norbert Hanke, CEO of Hexagon Manufacturing Intelligence. “It’s all about timing. They have all these different islands of information, but the question is always about how best to connect it.”

In this comment, Hanke hints at some of the issues that must be addressed before manufacturers can take full advantage of the benefits of the smart factory.

“You have raw data and to a certain extent even an overflow of data,” says Hanke. “But when you bring intelligence to it (let’s call it ‘know-how’), you make information, and relevant information is important. What is relevant? That is the question that leaders need to think through for their own processes.

“We’re getting better,” he continues, “but there’s still a long way to go.”

The obstacles

What Hanke refers to here is the distance yet remaining before industry might enjoy a truly smart, completely integrated factory that can control waste and increase value, both within additive and subtractive manufacturing processes. That means adjusting in real time to the countless dimensions, degrees, and conditions that occur in factories, and using everything that the information is saying (and some of what the information is not saying) to manufacture more efficiently. The feedback loops that Rollen mentioned in his keynote can only occur when machines achieve decision-making abilities far beyond anything possible today. To get there will require not just machine reason and machine logic, but also machine deduction and even a semblance of machine intuition.

All this looks great on paper (or, as the case may be, a CAD file), but there are major hurdles to be overcome before anything approaching real machine intelligence can be achieved. For an understanding of how far this technology has come, and how far it still has to go, consider one of Hexagon’s recent acquisition, MSC Software Corp.

During the Hexagon Manufacturing Intelligence divisional keynote on the second day of the show, Hanke turned over a portion of his presentation to Dominic Gallello, president and CEO of MSC, a simulation software company with a legendary history. MSC was founded in 1963, and just two years later won a bid from the fledgling National Aeronautics and Space Administration (NASA) to create a structural analysis program, NASA Structural Analysis (NASTRAN), to help simulate, analyze, and predict stress, strain, vibration, dynamics, and acoustics, as well as perform thermal analysis for NASA’s Gemini and Apollo programs. It was one of the tools that quite literally helped humans reach the moon.

Now MSC is lending its deep experience to another long-term project of fantastic implications and gargantuan financial import: The quest for autonomous vehicles. Like the space race of a half-century ago, the competition to perfect a self-driving car is now occupying the time and minds of some of the top engineers in the world. During the 1960s, landing a man on the moon and returning him safely to Earth primarily concerned nationalized efforts from the United States and the Soviet Union; today, autonomous driving is a focus of private and public enterprises in a variety of sectors as well as pretty much every major industrial center on the planet.

The key to autonomous driving, according to Gallello, is “event-space simulation.” When it comes to driving, that event space is extraordinarily complex. Think of a typical busy intersection and everything that is going on around the car within that maelstrom of data. Human drivers can immediately perceive enough of what’s happening and go or stop accordingly, based on experience and a sense of what other human actors will do. Or consider driving through a dark tunnel: You and I know that the road is there and clear even though we can’t see it, so we keep going. But how do you program machines to handle these tasks?

The sheer amount of data that need to be interpreted is almost incomprehensible; Gallello and his team talk in terms of exabytes of data. How big is an exabyte? It’s a quintillion bytes (a one followed by 18 zeros). You may prefer to think of an exabyte as a million terabytes. Long story short, it’s a lot. To make sense of that much data and turn them into information requires a neural network of the kind that doesn’t exist today—outside of the human brain, of course.

The automated decision-making apparatus of the smart factory pales in comparison to the complexity found in self-driving vehicles, but the one has significant effects on the other. The lessons learned on roads and tracks across the world will help shape next-gen manufacturing. Machine intelligence is the link that combines the two.

Further to that point, I had the opportunity during HxGN Live to record a video interview with Stephen Graham, vice president of marketing for Hexagon Manufacturing Intelligence. In our discussion, he revealed what MSC brings to the company, expanded on the digital thread of information that passes throughout smart factories, discussed the utility and meaning of information, and a lot more. You can watch the short segment below.

The reality

The great catchphrase you hear about manufacturing today is “Industry 4.0.” During his keynote, Hanke walked us through the progression of Industrial Revolutions that brought us here, from mechanization during the 18th and 19th centuries to mass production in the 20th century to computerization and automation in the late 20th and early 21st centuries. Now, in this fourth and ongoing revolution sweeping over industry, the great advances involve the internet of things, cloud computing, and the functional integration of discrete processes. Next may come optical quantum computing, in which the power and speed of processing will be unlike anything we’ve seen before. If Moore’s Law is to continue, such breakthroughs, which currently live mostly in the realm of theory, will be imperative.

Given the amount of information that smart factories produce and must consider, digital storage capacity and processing power will continue to be all-important. However, the “islands of information” that Hanke referred to must also be bridged by speed and efficiency. That’s a point that Rollen made during a press luncheon at the show.

“The trend of edge computing, very few people discuss today,” he said. “You hear cloud computing, you hear IoT, but how much do you really hear people discuss edge computing? And I think that’s where they go wrong.”

Rollen is keenly interested in edge computing because it optimizes the cloud in a way that perfectly suits many of the sectors and systems that Hexagon operates within—not only manufacturing, but also surveying and geospatial applications. In basic terms, computing at the edge means de-centralizing data processing and analysis by pushing it closer to the source of the collected data. The result is greatly increased speed and efficiency vs. standard networks. For manufacturing customers with multiple sites, as an example, edge computing has the practical appeal of greatly reducing the constraints of latency and the demand for bandwidth. Edge computing is a tool that makes factories not only smart, but nimble as well.

The sense of nimbleness and efficiency represented by edge computing is, in a broader context, a perfect metaphor for the “data vs. information” comparison made by Rollen during his keynote. Like raw data, standard networks offer potential value for those with time and patience; edge computing, like information, is targeted, direct, and immediate. In the fast-paced, ever-changing world of manufacturing embodied in Industry 4.0, the latter will beat the former every time.

As you may have gathered from this brief report, HxGN Live is a place where big ideas come out to play, and those ideas are presented in an almost matter-of-fact fashion by thought leaders such as Rollen, Hanke, and Graham. Smart factories, artificial intelligence, machine intuition, autonomous vehicles, and complex simulations were all discussed within the realm of science, not science fiction. It’s a conference that is, literally, engineered to address the possibilities of tomorrow given the practicalities of today.

It is also an overwhelming experience, in a good way. As an attendee returning to the conference after a break of a few years, I was amazed at the changes in the scope of Hexagon’s vision, but perhaps I shouldn’t have been so surprised. Rollen casually mentioned during his keynote that the human race has created more data since 2015 than was created in all of history up until that point. A few years in this industry of ours can create earth-shaking changes. With that in mind, it’s tantalizing to ponder what we’ll see from this event, and this company, in the years to come.


About The Author

Mike Richman’s picture

Mike Richman