By Alberto Moel | August 28, 2020

This is the second installment of a two part analysis focused on the connection between automation, manufacturing flexibility, and productivity. In Part 1 we looked at US Total Factor Productivity (TFP) in manufacturing using data from the US Bureau of Labor Statistics (BLS), and found a worrisome downward trend over the last couple of decades. After some digging around, we identified a broad shift toward capital and away from labor in manufacturing (i.e., lower labor intensity in production) as one of the culprits in the TFP slowdown.
This shift is indicative of higher automation levels. Too much or the wrong kind of automation causes inflexibility, which, in light of manufacturing trends that require more flexibility, causes productivity in the aggregate to suffer. To contribute to higher productivity (and higher TFP), it would be much better if automation worked in harmony with labor.
Now that we have provided macro evidence of the impact that inflexible automation has on productivity, let us do a light romp through the micro. Relying on a few examples, we will show how the inefficiencies and inflexibilities of high levels of automation show up in suboptimal labor and capital allocations and higher-than-warranted manufacturing costs. In particular, we will reiterate a point implicit in the definition of optimality: too little automation is also suboptimal, and the limits to automation are sometimes driven by technology evolution hitting a wall. In other words, there is (in theory and in practice) a “best,” most productive combination of humans and machines working together.
Too Much Automation
Industry-wide work done in the early 2000s by Frits K. Pil and John Paul MacDuffie as part of Phase Three of the International Motor Vehicle Project1 shows, in a roundabout way, that automation doesn’t necessarily mean higher productivity. This conclusion is more explicitly supported by Ron Harbour, a manufacturing expert at Oliver Wyman (and author of the highly regarded Harbour Report), who was quoted in 2017 saying:
“When it comes to what functions to automate, we’ve seen differing philosophies. The manufacturers we saw as most competitive—the Japanese, the Koreans—needed a business case for automation. Whereas automakers in Western Europe and even the United States went more aggressively towards automating things, even when that didn’t pay off. We’ve seen examples of companies installing automation that required more people, with higher skills, than were required before. They did it just to display their technological prowess. I’ve always been dismayed by that. Ironically, the most [heavily] automated factories in the Harbour Report are not in the top quartile [in the productivity ranking]. Many are in the bottom.”


Max Warburton and Toni Sacconaghi of Bernstein Research2 have used IMVP data and Ron Harbour’s quantitative analysis to illustrate an inverse correlation between automation levels and labor productivity. Figure 1 (adapted from Max and Toni’s report) shows hours per vehicle data for specific manufacturing plants by German, French, US, Korean, and Japanese car makers in the EU and US.




Depending on the relative labor and machine running costs, more hours per vehicle does not necessarily translate into higher costs per vehicle, but if the allocation of labor and capital were optimal, we would see both increasing automation and decreasing hours per vehicle. We do not, so it is straightforward to conclude that over-automation is counterproductive.


Figure 1: Hours per vehicle and automation across the automotive industry.





Too Little Automation
It is also the case that in many verticals and manufacturing steps, automation is just not present. Figure 2, collected by a different team at Bernstein Research from industry data, shows an estimate of the level of automation for a number of steps in various industry verticals. As we can see, there are great variations in the penetration of automation among different steps. Some are more or less fully automated, while in others automation is basically non-existent—and this is even within the same industry vertical.






The common driver of this disturbing heterogeneity is the level of human dexterity, judgment, and ingenuity that would need to be “solved” for with automation in a particular manufacturing step. Steps with greater needs for human skills would not only be too complex to design and build with automation, but also prohibitively costly to attempt to automate.




Figure 2: Automation levels in different industry verticals.

Just the Right Amount
Gorlach and Wessel3 examine VW plants in Wolfsburg, Germany, and Uitenhage, South Africa and find that there is an “optimum” level of automation for a given manufacturing process. Figure 3 (more or less adapted from their research) shows how a mix of labor and capital (i.e., automation and other hardware) leads to a U-shaped total cost curve, where costs are minimized for a particular combination of labor and capital.4


To ground this general concept, let’s review an application that we have written about in the past: palletizing. In a case study, we analyzed a palletizing process that requires an intermediate step where “bumpers” are mounted on the four corners of the pallet as the pallet is being “built up” by the robot. These bumpers are there to protect the boxes on the pallet (for example, of shampoo bottles) from damage as they are transported to a warehouse or point of sale.
The palletizer itself, a standard large-payload ABB Model IRB460 palletizing robot arm, was not designed to install the bumpers on its own, so the end user had to consider various solutions. One approach was to automate the whole palletizing process and build a specialized piece of automation—a “bumperizer”—with the sole purpose of mounting the four bumpers at the appropriate point in pallet construction. This approach entailed low labor content and high capital costs,5 falling on the left side of Figure 3.
Alternatively, the end user could go all-human, eliminating the palletizing robot and relying entirely on manual processes for pallet stacking and bumper mounting. This approach entailed high labor content and low capital costs, falling on the right side of Figure 3.
Neither solution allowed the manufacturer to keep the cost of palletizing at an optimum low. Only the third approach, where the robot did the palletizing and a human installed the bumpers at the appropriate time, kept labor and capital costs at levels that resulted in lower total unit costs. This approach—a combination of humans and machines working together—was superior to both fully automated and fully manual operation, but was only possible with a system that enables safe human-machine collaboration.





Figure 3: Representative cost structure in automotive manufacturing as a percent of labor content.




What Limits Humans and Machines Working Together?
It is clear from our current and past discussions that in some instances automation can introduce inflexibility in environments where it can destroy value or reduce productivity. In these situations, the ability to “dial back” some of the automation and replace it with human flexibility, dexterity, creativity, and judgment would be the right approach.
The limitation there is the fact that humans and machines cannot work safely in close proximity. Breaking that barrier will allow for value creation and productivity improvements as both capital and labor are utilized more efficiently.


The inefficiencies plaguing the manufacturing industry stem from technological limitations… when those technological barriers are overcome, it’s possible to unlock productivity gains with an optimal combination of humans and machines.


State of the art sensing, processing, and actuation technologies are hardly comparable to humans, and their limitations result in a need for human input in manufacturing steps, which in turn limits how much automation can be added to the process. All together, this leads to suboptimal allocations of labor and capital. A system like Veo FreeMove would allow for safe and enhanced human-machine interaction so that steps can incorporate optimal levels of automation.
What are the possible in-the-limit gains from an optimal mix of capital and labor in manufacturing? MIT’s Julie Shah6 developed a “human aware” robot execution system called Chaski that makes human-robot interactions more fluid. In her experiments, when a person collaborates with a Chaski-enabled Nexi, a dexterous mobile robot, to assemble simple structures using building blocks, the human’s idle time is reduced by 85% relative to all-human teams performing the same task.
Our own work at Veo finds about a 2x improvement in cycle time for a simple assembly task when a human is allowed to collaborate with a robot equipped with Veo FreeMove.
The inefficiencies plaguing the manufacturing industry stem from technological limitations — Professor Shah’s research and our case studies show that when those technological barriers are overcome, it’s possible to unlock productivity gains with an optimal combination of humans and machines. Until enabling technologies like FreeMove become widely available, productivity gains in manufacturing will continue to hit a wall and see only incremental improvements. What that “wall” looks like and how other technologies have gotten around walls will be the subject of our next blog post.


Editors Note: Robotics Business Review would like to thank Alberto Moel for permission to publish this piece (lightly edited). The original can be found HERE. All views, thoughts, and opinions expressed therein belong solely to the author.



Notes










2 In the interest of full disclosure, I used to work at Bernstein Research and I am “borrowing” one of their charts from their March 28, 2018 report, Tesla: Model 3 and the fallacy of automation–What we believe is wrong and why it may remain difficult to ramp production. The report is only available to Bernstein clients, and in it they basically hammer the point I am making in this short blog post that automation in capital goods manufacturing can have negative returns to scale.
3 In a paper coincidentally titled, Optimal Level of Automation in the Automotive Industry (Engineering Letters 16:1, 2008).
4 Three points to note. The “capital costs” in this chart correspond to the amortized or per-unit costs so as to determine an equivalence between the direct labor input and the capital cost per unit. Even for 100% labor content, capital costs are non-zero because fully-manual operations require capital expenditures for tools, fixtures, electronic controls, etc. And lastly, we make the implicit assumption that when unit costs are minimized, productivity is maximized. This sounds sensible, but it’s not a guarantee, although it’s pretty certain that minimizing costs results in improved (but not maximized) productivity.
5 And we mean high—a high-reliability bumperizer can cost $175,000, which is three times the cost of your average palletizing robot.
6 Fluid Coordination of Human-Robot Teams, MIT PhD thesis, February 2011.






Alberto Moel, VP of Partnerships and Strategy, Veo Robotics

Alberto Moel is the VP of Partnerships and Strategy at Veo Robotics, where he is responsible for industry partnerships, intellectual property, market and competitive analysis, and company strategy. Prior to Veo Robotics, Moel was a Senior Research Analyst at Sanford C. Bernstein in Hong Kong, where he covered Asian high technology companies in the automation, robotics, and manufacturing technology sectors.
Previously as a principal in the Hong Kong office and co-leader of the Tokyo office of Monitor Group, Moel specialized in technology strategy and corporate finance for Asian high-tech companies. In addition, he has been a Professor and Lecturer at the Hong Kong University of Science and Technology and at Harvard Business School, co-founded a successful Brazilian hedge fund, and worked at JP Morgan (New York and Mexico City), Toshiba Corporation in Japan, and IBM Corporation in New York, Cambridge, and Zurich. Moel has SB, SM, and ScD degrees from MIT, and an MBA from Harvard Business School.

Related Content: