Can AI Help Manufacturers Maintain High Quality Amid Uncertainty?

Drishti’s focus on enhancing employee capabilities is a fresh take on technological solutions

Published: Monday, January 24, 2022 - 13:03

We are a full two years into post-pandemic manufacturing life, with the omicron variant the latest cause for concern. It might sound hyperbolic, but I’m pretty sure manufacturing will never be the same as it was in 2019. In some ways, that’s a good thing.

One silver lining of the novel coronavirus was the adaptability that it forced on many manufacturers. For all the disaster planning and preparedness exercises companies practice, there’s nothing like a true crisis to test the system and show the gaps in it.

But pandemics aren’t the only hurdles on the manufacturing horizon. We know that natural disasters are increasing in frequency. Supply chain disruptions and political upheaval are also growing concerns around the globe. In a recent Drishti survey, we found that 60 percent of the 400 manufacturers that responded believed a significant disruptive incident would likely occur during the next 12 months.

With that level of uncertainty plaguing the industry, the question has to be asked: How do manufacturers continue to deliver high-quality products when they’re sure to be impacted by happenings outside of their control?

The key to staying adaptable: Humans in the plant

Although the buzz around Industry 4.0 is all about automation, connected factories, IoT, and the like, there isn’t a peep about humans in all that chatter. The fact remains that humans are the most adaptable beings on the planet. (OK, the Covid-19 virus and a few insect species might take umbrage with that claim, but the chances of them reading this article are slim.)

We saw this manifest in factories with sharp clarity during the early days of Covid-19. Illness, family obligations, and social distancing kept some workers away from the line, but the remaining line associates were able to adapt and modify their efforts to fill in the gaps. Can you imagine how long it would take to reprogram a line of robots to do double the work for which they were originally designed?

The manufacturers we talked to overwhelmingly saw the benefits of their human workforce. In fact, 73 percent agreed with the statement, “The more manual you are, the more adaptable you are.”

Adaptability doesn’t automatically mean high quality

Having human workers who can swiftly adapt to changes is a massive advantage when it comes to dealing with external threats. But it doesn’t automatically follow that being an adaptable company in the face of adversity also means you’re maintaining high-quality production standards, especially given the high attrition rates and daily absenteeism manufacturers are reporting. We wanted to better understand the internal challenges facing manufacturers as they battle outside threats.

When polled about the biggest challenge they face internally in the coming year, manufacturers said productivity, quality, and training were top of mind. It’s not surprising, because training is the very basis of good quality and high productivity.

It’s absolutely true that humans are more adaptable than automation. It’s also true that they are more variable in their processes. And the tenets of lean manufacturing are clear: Quality comes when processes are followed absolutely, every time. In fact, standardized work is the foundation of lean manufacturing: The implementation of one “golden way” of collaboratively designing and then completing an assembly task, consistently following the same order and length of time, every single cycle.

But standardized work adherence can be extremely difficult to monitor and enforce, and times of high uncertainty are particularly challenging. During May 2020 alone, manufacturers hired 639,000 new people. During the pandemic's peak months in spring 2020, 900,000 people left the manufacturing sector. That’s a huge number of fresh faces (7%, to be precise) starting at unfamiliar stations on assembly lines across the country. Traditional training programs—some as long as three months—and methodologies, from one-on-one master-apprentice scenarios to classrooms and dojo stations, struggle to bring that many people up to speed quickly. That’s when quality starts to suffer.

People plus new technology drive adaptability and quality

You can see why some manufacturers have blanched at the idea of doubling down on people, even when they know the adaptability of human workers makes their organizations more resilient to uncertainty and threats. Here’s where new technology comes into play, technology that quickly trains on and ensures standardized work adherence, even if an employee is still quite fresh to the line.

At Drishti, we’ve pioneered an extraordinary technology called action recognition. It works like object recognition—by using video and AI to identify objects within a frame—but it adds the dimension of time by analyzing video feeds as events unfold on the plant floor. So rather than noting a line associate holding a screwdriver—something object recognition technology can easily do—Drishti can tell you the line associate brought the heart monitor to his station, picked up a screwdriver, tightened the third screw, then rotated the monitor and tightened the first screw—in that order, with a time stamp for every step in the cycle that is defined by standardized work.

This level of cycle-time information fundamentally changes the quality game. All employees now have specific videos, tied to standardized work, to learn from—similar to athletes perfecting their performance by watching game footage.

• The line associate who sees these steps performed on a tablet in front of him in real time can respond to out-of-sequence alerts or missed steps before the unit leaves the station.
• The line supervisorr who has access to holistic shift data can begin to identify line associates who need additional training, or high performers who should be cross-trained on other stations for additional coverage on that station.
• The industrial engineer who is alerted to standardized work deviations can quickly identify which units were assembled by the deviant line associate and pull them for inspection before they leave the factory. Long-term, the engineer can devise and implement process changes that minimize these deviations.
• The plant manager who has this deep level of detail on every station and line in the plant can begin to fundamentally improve training, standardized work adherence, and quality across the entire plant floor.

I know this probably sounds like a bit of a sales pitch. But the truth is, without comprehensive, granular, and AI-powered ways to digitize human assembly actions, it’s difficult to be an adaptable manufacturer without introducing quality issues in your operations. That’s why I get so excited about what we’re doing—because with uncertainty now the standard state of manufacturing, plant personnel must figure out how to cope with that reality while continuing to produce quality goods for their consumers. We’ve found a way to assist in that pursuit.

About The Author

Prasad Akella

Prasad Akella is the founder and chairman of Drishti, which uses its patented action recognition technology to analyze video streams and digitize manual processes in assembly operations.