When the introduction of AI technology at Nichirei Foods' Funabashi Plant began to be considered in 2019, employees were skeptical about its feasibility. The Funabashi Plant manufactures a variety of rice products, including Nichirei Foods' flagship product, HONKAKUITAME-CHAHAN (fried rice).

Nichirei Foods has a long history of careful manufacturing with human eyes and hands. That is why many people were skeptical that an AI could perform as well as a human. Takei, who has been involved in rice product manufacturing for many years, is a line leader of the Funabashi Plant Production Group in the Production Management Department. He has experience in manufacturing departments at several plants in Japan and currently leads the rice production line, which has 170 employees.

Although we are working to automate and reduce manpower on the production line, there are still areas where we rely on human power, such as when trained employees check products on the line and manually remove burns. We can send out products with confidence only after confirming their safety and security with human eyes and hands.

On the other hand, the burden on employees was a source of concern.

In addition, training is also required for people to instantly judge and remove foreign substances such as burnt food. The introduction of AI was also expected to be a way to solve the difficulty of human resource development. First, cameras, which are indispensable for the introduction of AI, were used on a trial basis in a hot and humid factory to test whether they could be used to take pictures without problems over a long period of time. Then, teams were formed for each of the three inspection processes that place a heavy burden on employees, and studies began.

One process was to inspect whether or not the fried rice bit into the crimped part of the product package.
The second was to inspect the green onion as an ingredient to see if any foreign substances were contained.
The last process is to check and remove burnt fried rice.

"I was excited by the challenge," recalls Tabikawa, who, like Takei, was selected to be a member of the team that introduced the AI technology to check for and remove burnt fried rice. At that time, as a member of the engineering group, he was in charge of the maintenance and preservation of the machines in the Funabashi Plant, as well as the study of introducing new equipment.

"The use of AI technology was encouraged by society as a whole, and I was proud of being able to try equipment that had not yet been introduced in other factories. In addition to Mr. Takei, who has a thorough knowledge of the site, members with knowledge from various departments, such as quality assurance and equipment development staff, gathered together and began the challenge."

The goal was to create an AI device with eyes as good as or better than those of humans.

The new device uses AI to recognize scorch marks in the data captured by the camera, and then conveys the information to a device that removes the scorch marks. The first thing that was discussed was the location of the camera that recognizes burns and the device that removes them.

The development of the equipment itself was carried out in parallel, led by Tsutsui of the Equipment Development Group. The shape and size of the equipment were not determined until the very last minute, so the layout of the production line had to be conceived with a lot of imagination, which was also a challenge.
(For more details on the trial-and-error process leading up to the development of the device, please see Part 2: How AI Came to the Frontlines of Manufacturing )

Above all, we must avoid lowering the safety and hygiene level of our products.

The team members created more than 20 layout plans and examined them one by one to see which one would satisfy the requirements. After more than six months of weekly meetings, they finally found a layout that would allow for the installation of large equipment without causing safety problems. After enthusiastically explaining the layout to those in charge of each process and persuading them, a location was finally determined.

After the location is decided, the phase of refining the AI's identification performance begins. Here, Takei made a request from a manufacturing perspective. That was to make the AI capable of judging other things besides burnt food.

Therefore, we sent test products with uneven amounts of ingredients and let the AI learn from various types of abnormalities. Through trial and error by the members and the equipment development group, the AI's accuracy gradually improved, and they were able to establish an alert function.

Tabikawa hastened the verification process to ensure that information was transmitted smoothly from the camera to the device to be removed.

Nevertheless, we patiently continued the verification process while communicating with the persons in charge of each machine manufacturer. About eight months had passed since the study began. When the accuracy of the system became clear during the trial introduction, employees were surprised.

Employees previously involved in the inspection process are now able to work in a different process. Takei says the response has been positive, as the training period required to work in the inspection process has been eliminated.

Although the burn removal equipment is currently in actual use in the manufacturing process, Tabikawa is aiming for even greater heights.

The introduction of the burn removal device has led to the introduction of AI in other processes at the Funabashi Plant. Nichirei Foods will continue to create safe and secure workplaces and deliver the same great taste to our customers' tables.

For more information on the appeal of the HONKAKUITAME-CHAHAN (fried rice) manufactured at the Funabashi Plant introduced in this article, please refer to [Part 1: In-depth report by the fried rice fairy on the appeal of "Authentic Fried Rice ®︎"!