In preparation for mass customization, for starters, Japanese and German tech research officials committed to expanding their joint work to establish a “social-technical or maybe ‘cyber-social’ environment where ‘digital companions’ and production lines communicate with humans” working in manufacturing, Andreas Dengel said at Germany’s CeBIT fair in March.
“Industry 4.0 has many consequences, not only for environments like factories but also active components, activators agents and human beings,” said Dengel, who heads the smart data & knowledge services department of the German Research Center for Artificial Intelligence (DFKI). “A digital companion is someone who is looking over your shoulder and observing your behavioral aspects—who is looking at what is relevant in a certain situation and tries to overcome complexities.”
End goals include reducing data, process complexity and interactions—“to concentrate on the most relevant part of a job.”
This becomes critical in the face of mass customization.
“If you look at ‘batch size: one’ production, it means individuality in the future,” Dengel said. “It also means the processes for configuring and maintaining things will change, and this requires a new way of educating: One, it is insight [into issues that require machine learning uncovers best], and two, it is also some companion who is taking over some roles and completive aspects.”
DFKI worked on the technological innovation for the last year with Hitachi, which demonstrated it at CeBIT.
The pact signed in March broadens that work in Japan, starting with expanded use inside Hitachi, and makes it available to German firms that want insight into work activity.
Hitachi’s work looked at humans working on assembly lines, Hisashi Ikeda, GM of Hitachi’s Center for Technology Innovation-Systems Engineering, said in an interview with Smart Manufacturing magazine in Germany.
They were outfitted with eye trackers that measure visual attention and sensors on their arms that measure muscle contractions. “With this combination of body attention and visual attention, you can really control what’s going on, and the companion can monitor what you are doing and measure your workload or cognitive load and control whether you are doing the right things,” Dengel said.
The technology should help with the worldwide skilled labor problem in manufacturing, as well as productivity, Ikeda said. “Manufacturers have a problem with lost costs” that the technology can help address, he added.
All manufacturing segments that involve manual operations will benefit from the technology, Ikeda said.
“The technology is unlimited,” Dengel added. “Everywhere you use your body to do something, you can measure body activity and visual activity” and bring about improved processes.
Eye-tracking work pays off
DFKI came up with the idea for the social-technical environment inhabited by humans in manufacturing and their “digital companions,” he said.
Part of the effort grew out of DFKI’s work on eye tracking. “The eyes interface with the brain, and how you are looking and whether you are hesitating, whether you are interested—the eye muscles behave differently,” he said.
That can tell an assembly line manager a lot about a worker’s cognitive load. By studying eye behavior, one can spot motivation, interest, excitement, problem investigation and comprehension, Dengel said.
The work is likely to be used in the future to train robots, as well: After experts outfitted with cameras and eye triggers and muscle sensors show how a task is most efficiently performed, that intelligence can be used to program the bots “on what’s relevant in a certain situation,” Dengel said.
The work undertaken in Japan in the last year also incorporated what DFKI learned from working with graphic processor unit creator NVIDIA on a form of machine learning called “deep learning,” which enables what he described as super-human capability to recognize emotions in pictures and videos.
In manufacturing, deep learning is used along with cameras to check whether a worker is handling the correct object or not, Dengel noted.
Worth noting: The commands that result from all the number crunching are not 100% task oriented. The artificial intelligence is also used to measure fatigue—and advise workers when to take a rest.