The factory floor is a marvel of automation. With a press of a button, the whole place can seem to run itself. But although today’s factories use automated workflows, process change is still mostly manual. When demands arise in an industrial environment, managers and engineers must interrupt the automation to update the processes that make the machines go.
Now, thanks to machine learning algorithms, it’s becoming possible for smart software to scrutinize data from a variety of sources — sensors on machines or changes in supply chains, for instance — and redesign processes in real time.
In our survey (not yet publicly published) of almost 170 industrial organizations, 96% of respondents agreed or strongly agreed that machine learning is automating process-change management inside their organization. Overall, more than half of respondents directly attributed significant, often exponential improvements in business processes to machine-learning-enhanced processes.
Observing early adopters, we’ve found that automated process change comes in three main flavors: self-adapting, self-repairing, or a combination of the two. Some organizations are already deploying elements of process change automation, while others are developing technologies that lay the foundation for it. Below we present examples of companies leading the charge.
Automobile manufacturers know that customers are increasingly interested in customizing their cars. But how can processes adapt to enable cost-effective customization? Andreas Nettstäter, who leads European research projects at the Fraunhofer Institute of Material Flow and Logistics (IML), says a new Fraunhofer initiative called Smartface Consortium is tackling this challenge.
Smartface is testing embedded sensors and smart workstations in car-making plants to create a kind of self-adapting assembly line that can modify the steps in its process to fit the demands of various automotive features. Rather than an end-to-end assembly line, cars shuffle in optimized, nonlinear paths among reprogrammable stations that flexibly perform tasks for building a car to specification.
Soon, Nettstäter says, electric vehicles and other cars with numerous digital systems will be even more individualized than cars are today. “We need to get rid of the static production lines where only one car and one configuration can be built,” he says, adding that the automotive industry in Germany can already support over one million configurations of a car model — as opposed to U.S., Japanese, and Korean companies, which produce only three standard configurations.
In addition to customizability, another benefit of self-adapting assembly lines is robustness. “If one station has a failure or is broken down, the others could also do what should have been done in this assembly station,” Nettstäter says. Fraunhofer IML’s research confirms our survey results, in which 92% of respondents agreed or strongly agreed that machine-learning-enabled processes were able to optimize themselves without human intervention. Such self-adapting processes are driving a new breed of cost-effective assembly lines that meet consumer demands for highly customized products.
Our survey shows that 79% of respondents agree or strongly agree that machine-learning-enabled processes can perform a kind of self-repair in real time.
Chief Analytics Officer Europe
15% off with code 7WDCAO17
Chief Analytics Officer Spring 2017
15% off with code MP15
Big Data and Analytics for Healthcare Philadelphia
$200 off with code DATA200
10% off with code 7WDATASMX
Data Science Congress 2017
20% off with code 7wdata_DSC2017