Featured Product
This Week in Quality Digest Live
Innovation Features
Matt Fieldman
Finding the right kind of crazy
Julia Canale
Benefits and challenges of becoming an early adopter of the technology
John Colmers
Maryland’s innovative and sound approach could be the solution for rescuing systems nationwide
Christine Schaefer
‘Our Baldrige journey has enabled us to destroy many of the belief systems that keep governments from improving’
Alena Komaromi
Even zero-sum negotiations can turn into a win-win

More Features

Innovation News
Minimizes manual labor in high-risk roles associated with high turnover
Eiger Fleet to enable more control and automation of distributed manufacturing
The tabletop diagnostic yields results in an hour and can be programmed to detect variants of the SARS-CoV-2 virus
First Responder UAS Triple Challenge focuses on using optical sensors and data analysis to improve image detection and location
More than half of respondents expect to meet Industry 4.0 goals within two years
Both quality professionals and their business leaders agree that openness and communication is essential to moving forward
Voxel8 patented technology will provide printed lattice structures to be used as inserts in midsoles
Purpose-built for cannabis analysis
True 3D holographic displays are practical with only moderate computational requirements

More News

Ryan E. Day


A Human Touch in Smart Manufacturing

Sorting through the chaff of Industry 4.0 hyperbole

Published: Monday, October 1, 2018 - 11:02

The fourth industrial revolution is upon us. Collecting real-time production data is becoming more common as enterprise-wide software develops into a tool that enables transformative gains in productivity. Although not common, there are “lights-out” factories, such as the fully automated FANUC manufacturing facilities, in operation today. And AI has become a ubiquitous plug-in for nearly every sector imaginable. “Industry 4.0” and “smart manufacturing” are now familiar terms in the manufacturing lexicon. The caveat to the fanfare is the same as with the first industrial revolution: “What is the future of humans in all this?”

Man vs. robot misconception

Although automation is only one aspect of Industry 4.0, it seems to receive an inordinate share of digital ink, second only to AI. The nugget of concern seems to be fear of robots and full automation replacing human employees in manufacturing. Considering that the debate over the effect of current levels of automation on levels of unemployment still rages, perhaps it’s a concern worth examining.

Because there is so much media coverage of robotics advancement, it’s easy to jump to the conclusion that robots will soon be the major workforce in manufacturing. We can’t escape the nonstop videos of robots demonstrating new levels of dexterity and speed. But, is it actually possible that robots are the new perforce factory employee?

Dr. Prasad-Akella-Drishti
Dr. Prasad Akella, founder and CEO of Drishti

During the 1990s, Prasad Akella led General Motors’ initiatives to develop co-bots for the factory floor, and literally changed the face of shop-floor dynamics forever. With a Ph.D. in robotics from Stanford University, and decades of practical application of robotics, Akella has a deep understanding of the subject of robots in manufacturing.

Akella was kind enough to take time out of his busy schedule as founder and CEO of Drishti to answer a few questions for us.

Quality Digest: It seems everywhere we turn, we see news about automation, robots, and AI. Your view seems very human-centric by comparison. Aren’t we looking at a future of more and better robots?

Prasad Akella: The headlines are scary, aren’t they? “Robots are coming for all of our jobs!” That’s what it sounds like. Especially in manufacturing, but the headlines don’t reflect the reality.

The truth is that up to 90 percent of manufacturing tasks are still performed by humans. Further, I believe humans will remain dominant, even in the face of rapidly advancing AI and robotics. And there are three reasons why.

The first is that there won’t be enough robots to replace humans for decades. It’s estimated that one new robot replaces 5.6 workers, but there are still today 135 people for every robot. The global industrial robot population is only expected to increase by 1.7 million through 2020. This won’t meaningfully displace the more than 340 million people who work in factories globally.

Second, manufacturing trends are on humanity’s side. For example, the mass customization trend (e.g., every consumer wants a product customized for his/her specific needs) is reshaping manufacturing around “lot sizes of one,” increased product options, and shorter times from order to delivery. There’s a reason that companies like Foxconn rely so much on manual labor: Robots can’t adjust to changing market demands nearly as fast as human workers.

But, even if robots were more agile, there still aren’t enough robot specialists to go around. This is the third reason. Every robot requires an ecosystem of programmers, process engineers, and skilled technicians. The skills gap and labor shortage may limit the expansion of robots as much as the capital requirements.

QD: Are there areas where humans are a more advantageous choice?

PA: Absolutely. Let’s talk about training and coping with process changes. Robots need an ecosystem to learn: specifically, they need skilled engineers, programmers, and process designers, all of whom are in short supply right now. You and I, on the other hand, learn very quickly and can adapt to new processes or unexpected circumstances without missing a beat.

Similarly, humans are much more flexible than robots. Consider robot “hands.” (I did my doctoral thesis on this subject more than 25 years ago!) Robots and automated equipment handle physical materials using what are technically called an “end effector.” There are relatively few general-purpose end effectors in the plant today—and certainly very few that are soft, controllable, yet capable of dexterously manipulating objects, large and small. Most parts and processes require custom-built end effectors. When the process or parts change, the end effectors need to change, too. But you and I, we can just readjust our grip.

Finally, humans are much easier to scale. More than 340 million people work globally in manufacturing, as compared to only approximately 2 million industrial robots. If you’re following what’s been happening at Tesla, this is crystal clear: Robotic lines are notoriously hard to design, install, scale, and run. Meanwhile, government data tell us that there are 141,000 manufacturing workers in the region around Tesla’s Fremont factory, any one of whom could show up at Tesla’s doors tomorrow morning.

QD: Industry 4.0 seems to be digitally connected assets—that is, machines connected to data systems. How do human workers fit into that?

PA: Industry 4.0 is about data. Specifically, it’s about finding opportunity hidden in massive datasets. Up until now, the only source of massive datasets has been machines, because it’s very easy to pull data off machines. Industry 4.0 has ignored people because it’s really hard to pull data from human activities at scale.

This means that people are Industry 4.0’s biggest blind spot. If up to 90 percent of the value creation and variability is coming from the human workforce, you can get as much data as you want from the 10 percent of machine tasks, but the insights you uncover are unlikely to address the most significant problems and opportunities.

Most directly, the Industry 4.0 team likely assumed that it would be impossible to measure human-driven processes, since it is extremely hard and has been an unsolved problem for a hundred years. Drishti is proud to be adding a whole new dimension to the Industry 4.0 discussion—arguably far more important than the machine discussion.

QD: What does the Drishti system provide that is missing?

PA: The most common way to get data from human activities hasn’t changed in about 100 years. I’m talking about time and motion studies. These were pioneered a century ago by Frederick Taylor and Frank and Lillian Gilbreth, around the time of Henry Ford. The key tool was a stopwatch and a spreadsheet.

Today, when I walk the plant, I see engineers standing around performing time and motion studies using laptops to type and smartphones to time—but it’s the exact same technique. Time and motion studies remain manual and subjective. You don’t get a big enough dataset, and the data you do get are biased by the fact that the person you’re watching knows that you’re watching, and is likely to act differently than if you weren’t there.

The dataset on human-driven process activities is glaringly insufficient when compared to the criticality of the decisions these data are used for—from daily staffing to capacity planning to setting prices for the products a manufacturer sells, and much more. The plant is making these decisions more or less blind.

Drishti’s system solves this problem by creating data from every human-driven process step or action it observes. For the first time, a plant can have continuous, system-wide analytics from its human workforce.

This is noteworthy for two reasons: First, Drishti is creating a brand new dataset derived from human-driven activities; and second, it is deploying AI to collaborate with and augment humans on the factory floor—as opposed to displacing them.

QD: Are we talking about a massive upgrade to classic scientific management methods of improving workflows and labor productivity?

PA: Yes. When you have a dataset this rich, the classic methods and principles get far more relevant. At the same time, there are unbelievable use cases for these data that scientific management could never have anticipated. Our customers come back to us every day with new applications for this dataset.

We already have some very interesting data on human effectiveness in a manufacturing setting. These data have the potential to help us rethink our work and lives. Very profound changes are in the offing. In the future, I fully anticipate an academic discipline to emerge: the science of analyzing physical actions at scale.

QD: If human labor can be significantly maximized by Drishti’s systems, what does that mean for manufacturing productivity?

PA: Rather than “maximized,” I’d frame it this way: For years, manufacturers have been making far-reaching decisions with the assumptions that human quality and productivity have reached a ceiling. The question I’d ask them is: What would change if this ceiling could be lifted? How would it change the ROI of automation? Wouldn’t it make any individual worker suddenly far more valuable? And, harder to justify the robotic invaders?

There are two relevant parts to Drishti in this regard. The first is the analytics tools, which is used by everyone in the ecosystem (engineers, supervisors, plant managers) to find new ways to improve the system.

The second is the human/AI collaboration that benefits the operators at the station level. Drishti basically acts as operators’ second brain and third eye. It helps them remember what they’re supposed to do, and it helps them spot errors to ensure they do it correctly. It’s our judo move on robotics: using the technologies that are helping robots advance to actually make people more competitive against automation.

QD: Are there even further-reaching ramifications of that level of productivity?

PA: Drishti represents true digital transformation: taking something that was heretofore unmeasurable and bringing it into the digital world. If it’s truly the disruptive technology we hope it to be, then you’ll see the most forward-looking factories realigning their entire ecosystem around this new dataset.

Think of how ERP systems transformed the industry 20 years ago. They made basic financial and transactional data relevant to almost every facet of the business. And massive, global businesses now live and die (and organize themselves) around these data.

A similar potential exists for Drishti. Think of it this way: What is a factory? In the most abstract formulation, a factory is a place where physical operations are performed to add value to raw materials. For a manufacturer, physical activities are the entire reason they exist. If, suddenly, they have a complete and accurate dataset of every activity that’s relevant to their core reason for being, then yes, I’d safely say the ramifications go beyond productivity. Digital transformation at its purest.


About The Author

Ryan E. Day’s picture

Ryan E. Day

Ryan E. Day is Quality Digest’s project manager and senior editor for solution-based reporting, which brings together those seeking business improvement solutions and solution providers. Day has spent the last decade researching and interviewing top business leaders and continuous improvement experts at companies like Sakor, Ford, Merchandize Liquidators, Olympus, 3D Systems, Hexagon, Intertek, InfinityQS, Johnson Controls, FARO, and Eckel Industries. Most of his reporting is done with the help of his 20 lb tabby cat at his side.