The inefficiencies plaguing the manufacturing trade stem from technological limitations of automation. When these technological obstacles are overcome, it’s attainable to unlock productiveness features with an optimum mixture of people and machines.
That is the second installment of a two half evaluation targeted on the connection between automation, manufacturing flexibility, and productiveness. In Half 1 we checked out US Complete Issue Productiveness (TFP) in manufacturing utilizing information from the US Bureau of Labor Statistics (BLS), and located a worrisome downward pattern over the past couple of a long time. After some digging round, we recognized a broad shift towards capital and away from labor in manufacturing (i.e., decrease labor depth in manufacturing) as one of many culprits within the TFP slowdown.
This shift is indicative of upper automation ranges. An excessive amount of or the improper type of automation causes inflexibility, which, in mild of manufacturing tendencies that require extra flexibility, causes productiveness within the mixture to undergo. To contribute to larger productiveness (and better TFP), it could be a lot better if automation labored in concord with labor.
Now that we have now supplied macro proof of the affect that rigid automation has on productiveness, allow us to do a light-weight romp by means of the micro. Counting on a couple of examples, we’ll present how the inefficiencies and inflexibilities of excessive ranges of automation present up in suboptimal labor and capital allocations and higher-than-warranted manufacturing prices. Specifically, we’ll reiterate a degree implicit within the definition of optimality: too little automation can also be suboptimal, and the boundaries to automation are typically pushed by expertise evolution hitting a wall. In different phrases, there’s (in principle and in follow) a “greatest,” most efficient mixture of people and machines working collectively.
Too A lot Automation
Business-wide work accomplished within the early 2000s by Frits Ok. Pil and John Paul MacDuffie as a part of Section Three of the Worldwide Motor Automobile Project1 reveals, in a roundabout means, that automation doesn’t essentially imply larger productiveness. This conclusion is extra explicitly supported by Ron Harbour, a producing professional at Oliver Wyman (and writer of the extremely regarded Harbour Report), who was quoted in 2017 saying:
“In the case of what features to automate, we’ve seen differing philosophies. The producers we noticed as best—the Japanese, the Koreans—wanted a enterprise case for automation. Whereas automakers in Western Europe and even the USA went extra aggressively in the direction of automating issues, even when that didn’t repay. We’ve seen examples of firms putting in automation that required extra individuals, with larger abilities, than have been required earlier than. They did it simply to show their technological prowess. I’ve all the time been dismayed by that. Mockingly, probably the most [heavily] automated factories within the Harbour Report will not be within the high quartile [in the productivity ranking]. Many are within the backside.”
Max Warburton and Toni Sacconaghi of Bernstein Research2 have used IMVP information and Ron Harbour’s quantitative evaluation for example an inverse correlation between automation ranges and labor productiveness. Determine 1 (tailored from Max and Toni’s report) reveals hours per automobile information for particular manufacturing crops by German, French, US, Korean, and Japanese automobile makers within the EU and US.
Relying on the relative labor and machine working prices, extra hours per automobile doesn’t essentially translate into larger prices per automobile, but when the allocation of labor and capital have been optimum, we might see each rising automation and lowering hours per automobile. We don’t, so it’s simple to conclude that over-automation is counterproductive.
Determine 1: Hours per automobile and automation throughout the automotive trade.
Too Little Automation
Additionally it is the case that in lots of verticals and manufacturing steps, automation is simply not current. Determine 2, collected by a unique workforce at Bernstein Analysis from trade information, reveals an estimate of the extent of automation for numerous steps in numerous trade verticals. As we will see, there are nice variations within the penetration of automation amongst completely different steps. Some are roughly totally automated, whereas in others automation is principally non-existent—and that is even throughout the identical trade vertical.
The frequent driver of this disturbing heterogeneity is the extent of human dexterity, judgment, and ingenuity that may have to be “solved” for with automation in a selected manufacturing step. Steps with higher wants for human abilities wouldn’t solely be too advanced to design and construct with automation, but additionally prohibitively pricey to aim to automate.
Determine 2: Automation ranges in numerous trade verticals.
Simply the Proper Quantity
Gorlach and Wessel3 look at VW crops in Wolfsburg, Germany, and Uitenhage, South Africa and discover that there’s an “optimum” stage of automation for a given manufacturing course of. Determine three (roughly tailored from their analysis) reveals how a mixture of labor and capital (i.e., automation and different ) results in a U-shaped complete price curve, the place prices are minimized for a selected mixture of labor and capital.four
To floor this common idea, let’s evaluate an utility that we have now written about previously: palletizing. In a case examine, we analyzed a palletizing course of that requires an intermediate step the place “bumpers” are mounted on the 4 corners of the pallet because the pallet is being “constructed up” by the robotic. These bumpers are there to guard the containers on the pallet (for instance, of shampoo bottles) from injury as they’re transported to a warehouse or level of sale.
The palletizer itself, an ordinary large-payload ABB Mannequin IRB460 palletizing robotic arm, was not designed to put in the bumpers by itself, so the top consumer needed to take into account numerous options. One method was to automate the entire palletizing course of and construct a specialised piece of automation—a “bumperizer”—with the only goal of mounting the 4 bumpers on the applicable level in pallet development. This method entailed low labor content material and excessive capital prices,5 falling on the left facet of Determine three.
Alternatively, the top consumer may go all-human, eliminating the palletizing robotic and relying fully on guide processes for pallet stacking and bumper mounting. This method entailed excessive labor content material and low capital prices, falling on the fitting facet of Determine three.
Neither resolution allowed the producer to maintain the price of palletizing at an optimum low. Solely the third method, the place the robotic did the palletizing and a human put in the bumpers on the applicable time, stored labor and capital prices at ranges that resulted in decrease complete unit prices. This method—a mix of people and machines working collectively—was superior to each totally automated and totally guide operation, however was solely attainable with a system that allows secure human-machine collaboration.
Determine three: Consultant price construction in automotive manufacturing as a % of labor content material.
What Limits People and Machines Working Collectively?
It’s clear from our present and previous discussions that in some cases automation can introduce inflexibility in environments the place it could possibly destroy worth or scale back productiveness. In these conditions, the flexibility to “dial again” a few of the automation and substitute it with human flexibility, dexterity, creativity, and judgment could be the fitting method.
The limitation there’s the truth that people and machines can not work safely in shut proximity. Breaking that barrier will enable for worth creation and productiveness enhancements as each capital and labor are utilized extra effectively.
The inefficiencies plaguing the manufacturing trade stem from technological limitations… when these technological obstacles are overcome, it’s attainable to unlock productiveness features with an optimum mixture of people and machines.
Cutting-edge sensing, processing, and actuation applied sciences are hardly comparable to people, and their limitations lead to a necessity for human enter in manufacturing steps, which in flip limits how a lot automation might be added to the method. All collectively, this results in suboptimal allocations of labor and capital. A system like Veo FreeMove would enable for secure and enhanced human-machine interplay in order that steps can incorporate optimum ranges of automation.
What are the attainable in-the-limit features from an optimum mixture of capital and labor in manufacturing? MIT’s Julie Shah6 developed a “human conscious” robotic execution system referred to as Chaski that makes human-robot interactions extra fluid. In her experiments, when an individual collaborates with a Chaski-enabled Nexi, a dexterous cellular robotic, to assemble easy buildings utilizing constructing blocks, the human’s idle time is lowered by 85% relative to all-human groups performing the identical process.
Our personal work at Veo finds a few 2x enchancment in cycle time for a easy meeting process when a human is allowed to collaborate with a robotic geared up with Veo FreeMove.
The inefficiencies plaguing the manufacturing trade stem from technological limitations — Professor Shah’s analysis and our case research present that when these technological obstacles are overcome, it’s attainable to unlock productiveness features with an optimum mixture of people and machines. Till enabling applied sciences like FreeMove turn out to be broadly obtainable, productiveness features in manufacturing will proceed to hit a wall and see solely incremental enhancements. What that “wall” seems to be like and the way different applied sciences have gotten round partitions would be the topic of our subsequent weblog submit.
Editors Observe: Robotics Enterprise Evaluate want to thank Alberto Moel for permission to publish this piece (flippantly edited). The unique might be discovered HERE. All views, ideas, and opinions expressed therein belong solely to the writer.
2 Within the curiosity of full disclosure, I used to work at Bernstein Analysis and I’m “borrowing” one in every of their charts from their March 28, 2018 report, Tesla: Mannequin three and the fallacy of automation–What we imagine is improper and why it might stay troublesome to ramp manufacturing. The report is barely obtainable to Bernstein purchasers, and in it they principally hammer the purpose I’m making on this quick weblog submit that automation in capital items manufacturing can have unfavorable returns to scale.
three In a paper coincidentally titled, Optimum Degree of Automation within the Automotive Business (Engineering Letters 16:1, 2008).
four Three factors to notice. The “capital prices” on this chart correspond to the amortized or per-unit prices in order to find out an equivalence between the direct labor enter and the capital price per unit. Even for 100% labor content material, capital prices are non-zero as a result of fully-manual operations require capital expenditures for instruments, fixtures, digital controls, and so on. And lastly, we make the implicit assumption that when unit prices are minimized, productiveness is maximized. This sounds smart, however it’s not a assure, though it’s fairly sure that minimizing prices leads to improved (however not maximized) productiveness.
5 And we imply excessive—a high-reliability bumperizer can price $175,000, which is thrice the price of your common palletizing robotic.
6 Fluid Coordination of Human-Robotic Groups, MIT PhD thesis, February 2011.
Alberto Moel is the VP of Partnerships and Technique at Veo Robotics, the place he’s accountable for trade partnerships, mental property, market and aggressive evaluation, and firm technique. Previous to Veo Robotics, Moel was a Senior Analysis Analyst at Sanford C. Bernstein in Hong Kong, the place he lined Asian excessive expertise firms within the automation, robotics, and manufacturing expertise sectors.
Beforehand as a principal within the Hong Kong workplace and co-leader of the Tokyo workplace of Monitor Group, Moel specialised in expertise technique and company finance for Asian high-tech firms. As well as, he has been a Professor and Lecturer on the Hong Kong College of Science and Expertise and at Harvard Enterprise College, co-founded a profitable Brazilian hedge fund, and labored at JP Morgan (New York and Mexico Metropolis), Toshiba Company in Japan, and IBM Company in New York, Cambridge, and Zurich. Moel has SB, SM, and ScD levels from MIT, and an MBA from Harvard Enterprise College.