(Reuters) — At a factory south of Japan’s Toyota City, robots have started sharing the work of quality-control inspectors as the pandemic accelerates a shift from Toyota’s vaunted “go-and-see” system, which helped revolutionize mass production in the 20th century. Inside the auto parts plant of Musashi Seimitsu, a robotic arm picks up and spins a bevel gear, scanning its teeth against a light in search of surface flaws. The inspection takes about two seconds — similar to that of highly trained employees who check around 1,000 units per shift.
“Inspecting 1,000 copies of the same thing day in and out requires a lot of skill and experience, but it’s not very creative,” CHIEF executive Hiroshi Otsuka told Reuters.”We would like to get workers out of these tasks.”
Global brands have long used robots in production, leaving the thorny paints of defect detection mainly to humans, but social estrangement measures to prevent the spread of COVID-19 caused a rethinking of the plant, which has stimulated the increased use of robots and other quality technologies, adding remote monitoring, which was already followed before the pandemic.
In Japan, such approaches constitute a sharp break from the go-and-see “genchi genbutsu” method that was developed as a component of Toyota’s production formula and adopted by Japanese brands for decades with almost devoted zeal. This procedure makes the staff responsible with constant monitoring of all facets of the production line to detect irregularities and has made quality control one of the last human strongholds in automated factories in a different way.
However, even in Toyota itself, when asked about automating more genchi genbutsu procedures, one spokesman said, “We are still looking for tactics in our production processes, adding automation processes where it makes sense to do so.”
Innovations in artificial intelligence (AI) have gone hand in hand with affordable equipment, but also with stricter quality needs in the customer component.
“We are seeing a gap between the quality of products manufactured on normal production lines and the quality demanded through our customers,” said Kazutaka Nagaoka, production director of Japan Display, an Apple supplier, as well as many automakers.
“The quality of products produced in automated lines is incredibly superior and more consistent,” Nagaoka said.
However, automating inspections is a challenge, given the desire to teach robots to identify tens of thousands of imaginable flaws for an express product and apply that learning instantly.Musashi Seimitsu’s low default rate of 50,000 sets left the company without enough faulty examples to expand an effective set of AI rules.But one solution came from Israeli entrepreneur Ran Poliakine, who implemented the generation of AI and optics he had used in medical diagnostics on the production line.His concept was to teach the device to detect the good, rather than the bad, to base the set of rules on up to a hundred very better or almost better sets, an amendment of the so-called gold sample.
“If you look at human tissue, you teach a set of rules about what’s smart and what isn’t, and you only have a moment to make the diagnosis,” he said.
Since this advance, the start-up of Poliakine, SixAI and Musashi Seimitsu, have created MusashiAI, a joint venture that develops and rents quality robots, a novelty in the field.
Consultations from automakers, spare parts suppliers and other corporations in Japan, India, the United States and Europe have quadrupled since March, when the new global coronavirus, Poliakine said.
“COVID-19 has accelerated movement. Everything is now on steroids because fleeing house displays, remote paintings can paint,” he said.
Earlier this year, auto parts manufacturer Marelli, which has an operational headquarters in Japan and Italy, also launched AI quality inspection robots at a plant in Japan, and the company told Reuters last month that it was looking for AI to play a greater role in quality inspections.. In the next few years.
Printer manufacturer Ricoh plans to automate all battery and toner cartridge production processes at one of its Japanese plants until March 2023.Robots are already performing maximum processes and, since April, technicians have been tracking appliances at the factory from home.
“Of course, you want to be available to compare and execute responses when disorders arise, but identity and confirmation are responsibilities that we can now carry out from home,” said Kazuhiro Kanno, general manager of Ricoh’s printer production unit.
Musashi Seimitsu might not say when he plans to have his plants fully automated, however, Otsuka said that AI complements, not threatening, the go-and-see system.
“The AI does not ask, “Why? Why?” But humans do.We hope to release them to ask why and how the defects occur,” he said.”This will permit more other folks to glance for tactics to consistently production, which is the purpose of ‘genchi genbutsu’.”
(Reports through Naomi Tajitsu and Makiko Yamazaki, additional reports through Maki Shiraki and Noriyuki Hirata, edited through David Dolan and Christopher Cushing).