On May 21, Jeff Moad wrote an article entitled, Should Manufacturers Video Plant Floor Performance? In it, Jeff described the technology coming out of my company, Drishti, which uses video-based systems to observe human activities for automatic validation and measurement. Think of it as an automatic time study performed on every cycle.

As Jeff rightfully pointed out, within most plants, the vast majority of value creation and the vast majority of variability are both traceable to human sources. And, as he further pointed out, it’s in everyone’s interest—management and labor alike—to simultaneously improve productivity, quality and safety.

We believe that Drishti is among the first of what will surely be many companies using video to help manufacturers observe and quantify processes that have been heretofore unmeasurable and, thus, invisible to analytics.

I also believe that video-based tools have the potential for the sweeping improvements that Jeff speculates. In any factory, there are countless issues that arise due to combinations of product design, process design and operator factors. Often these problems are clear to the human eye, if only there were enough humans to observe the system at all times. (Of course, there never are.)

Video-based computer vision will be used by manufacturers to direct a factory’s scarcest resource—human attention—to its most pressing issues as well as to generate descriptive statistics from the videostream that lets people trace the issue back to the root cause and, ultimately, solve the problem.

I also fully agree with Jeff on one other key point: introducing a video camera is not just a technical challenge, but a cultural one.

In particular, Jeff raised three very salient points about the ramifications of the camera as an input device. His points are well-conceived and I eagerly accepted his offer to respond in this forum.

The first point he raised:

“Plant workers and their front-line supervisors may well object to having their every move on the job recorded and analyzed.”

This would be the case, if not for one condition: Those working under the cameras benefit from the process improvement driven by a video system, too.

This is a critical point. Operators read the news. They see the headlines about the inexorable march of robots and AI into the factory. They worry about the long-term viability of their jobs. It’s understandable that they might be suspicious of video.

What convinces an operator otherwise is the understanding that a video system takes the technologies typically used for displacing people (computer vision, machine learning) and applies them to enhance humans, not to surveil them. It’s actually providing in-station operator assistance as the next generation of poka yoke: a judo move on robotics that lets human operators leverage technology that originated with robots as their “second brain and third eye.”

It is also the natural extension of what Toyota Production System developers Taiichi Ohno and Eiji Toyoda recognized: One has to empower the line operator with the tools and the authority to act on their insights in order to identify and solve issues.

Consequently, whenever Drishti begins an engagement, we insist that line operators are a part of the initial design team. They help us understand the challenges on the line, and we help them understand the purpose of the cameras. A well-designed video-based system, used appropriately, can help humans break the ceilings in productivity and quality that managers assumed were immutable and thus change the ROI calculation of an investment in machines—in the human’s favor.

Once operators understand this, we often see them engaging with the system as much, or even more, than the engineers behind the scenes.

Jeff’s second point:

“Manufacturers that turn to technology such as Drishti’s will need to reassure operators that the purpose is not to identify and get rid of poor performers‚ or those who may just be having a bad day—but to improve process performance overall”

I couldn’t agree more. The true purpose of Drishti’s system is make these human-powered systems competitive by marrying man and machine in three powerful ways.

  1. To help each operator break through perceived quality and productivity limits;
  2. To help the business identify the star performers in their midst who currently work diligently but with little measurable evidence of their work ethic or potential for advancement; and
  3. To identify specific opportunities for individual improvement that, with a little spot training, can turn a poor performer into a superstar.

Manufacturing is currently facing a labor shortage. The last thing any supervisor wants to do is fire an under-performer. The better course of action—for everyone—is to surface the path to improvement. And to retain these operators who understand the processes well.

Finally, his third point:

“At a time when many manufacturers are working hard to increase engagement with shop floor workers to drive cycle time improvement and reduce waste, the last thing they should want to do is to send the message that workers are not trusted.”

I’ll propose an alternate interpretation: The operators typically know the process far better than even the engineers who designed it. A truly collaborative system, video-based or otherwise, is designed to surface the wisdom these operators possess and, for the first time, give them the tools to prove the value of their ideas.

We call these the “brilliant outliers,” and they are perfectly capable of identifying faster, better, and safer ways of doing their jobs.

With video-based  systems, operators will now possess statistical evidence to have their experience recognized and their ideas heard. We believe this should deepen the trust between the company and its workers.

In closing, we believe that video will enable the two key organizational features of a true lean plant as defined by Womack, Jones, and Roos in their iconic book, The Machine that Changed the World: Transferring responsibility to the workers actually creates value, and enables a mechanism for identifying defects and tracing them back to their root cause.

Video is the means; lean is the ends.

 

Prasad Akella led the industry/university team that built the world’s first collaborative robots at GM (“cobots”, projected to be $12B market by 2025). He’s the founder and CEO of Drishti, a company deploying AI to collaborate with and enhance humans on the factory floor.