Report from the site. What I think is necessary to utilize AI / IoT in the manufacturing industry

I will post Qiita for the first time. I'm always indebted to you, but I'm sorry I didn't post it.

Introduction

As it is said to be the "cliff of 2025," digital transformation has become an urgent issue for the manufacturing industry. As emerging countries pursue quality and cost, you feel the rigors of Japan at the factory. I'm already wondering, "Is Japanese quality really good?" (Maybe it's just our factory!). However, it is clear that Japan needs to manufacture high value-added products for Japan, where the population is declining, and I think AI / IoT in particular is a major touchstone. I would like to convey the feeling of the site from the person who is in charge of AI / IoT (smiley w) at one maker.

I would like to say the following three points. ・ It is useless to create an AI / IoT promotion section only in the headquarters organization ・ Shopping alone is not enough, it is necessary to raise the overall programming level ・ Existing IT personnel may become obstacles

It is useless to create an AI / IoT promotion section only in the headquarters organization

Many large companies set up specialized departments for AI / IoT in a headquarters organization that is separate from the factory. I think this is a pattern that will not work if it is made only at the headquarters. I think that what is important in utilizing AI is high-quality data and software customized for the site. In that respect, I think it is extremely difficult for people at the headquarters to collect new data without knowing the existing data of the factory. In addition, even if they start collecting new data, they cannot deal with the troubles that occur every day immediately and leave it to the factory. Headquarters "I can't analyze because good data doesn't come up" Factory "I don't think it's hard to get data!" Is likely to be. After all, it is not enough to set up a promotion group only at the headquarters, and it is necessary to have a main team in charge at the factory. I think it's good to set up a group at the headquarters, but I think it's impossible for the headquarters alone to work.

Shopping alone is not enough, it is necessary to raise the overall programming level

I think there are many things that are often done by shopping. It's Tableau, it's DataRobot, and it's a pattern that I'm satisfied with "I'm doing AI / IoT" by shopping expensively such as DataSpider and SensorCorpus. I think this is really a waste in the sense that you can't master it. It needs to be customized to be used in daily manufacturing, and you have to write your own code somewhere. I think you can buy a tool or lose one hand without understanding this need.

Existing IT personnel can be a barrier

I want to say this the best of the three. It's been almost 30 years since the factory was built during the bubble era. Our factory is about that, but what is common in these factories is the legacy of the system. For example, my factory uses Oracle, but when I connect to the database with SQL, I get angry. Existing IT staff "What if I connect and become unstable?" "..." Why do you want to connect to Oracle with a command and issue a select statement with SQL to get angry (you get the data by clicking with the GUI software 30 years ago). .. AI development is unreasonable in such a state. Even if I only issue the select statement after a rush, if there is something, all the new code will be made into a scapegoat, so it's a daily affair. I'm full of apologies for writing a bat file that connects to Oracle, but I'm afraid. .. You're doing something wrong! ?? In addition, the bad news is that the knowledge of existing personnel is a generation ago. Even if you have a problem and explain "I don't understand Python !!" I completely lose my motivation.

Countermeasures

Based on the above, the necessary measures I think to utilize AI / IoT in the manufacturing industry are listed below.

It is useless to create an AI / IoT promotion section only at the headquarters

→ The headquarters is very helpful. But I think the factory also needs a team.

  1. In charge of infrastructure to ensure data quality,
  2. Backend manager who manages database and API
  3. And the front end person in charge of customizing according to the work of production engineering Is required. It may be okay to leave the development of algorithms that can be standardized to the headquarters, but I think that the main team is needed in the factory.
Existing IT personnel can be rather annoying

→ This is tight. To be clear, I'm tired. To avoid trouble, it would be ideal if the data line, database, and server were completely separated for AI / IoT. The database is set up with MySQL, the server can be managed with a personal computer using GCP and AWS, and the data line is also divided by the second ether. Even so, if there is a lot of data traffic and other software is said to have stopped, I laugh in my heart and cry.

It is necessary to raise the overall level

Everyone needs an understanding of programming. It's hard to take root among people who don't think they're involved, but I think the staff needs some understanding. I think it would be good to have students study under the name of "a tool that automates fixed procedures."

Finally

I think it would be great if I could master AI / IoT, but there are many issues in the Japanese manufacturing industry. It's old and sticks to something. .. It's easy to remake it from the factory! What? Ah, emerging countries have already won. ... No no! !!

I am developing an app with Angular. It is a business application that acquires text information from photos and cites it for use in blogs etc. at super speed. Apple [https://apps.apple.com/app/id1497498494][2] [2]:https://apps.apple.com/app/id1497498494

Google play [https://play.google.com/store/apps/details?id=com.rainbowsv2.ocr][3] [3]:https://play.google.com/store/apps/details?id=com.rainbowsv2.ocr

Recommended Posts

Report from the site. What I think is necessary to utilize AI / IoT in the manufacturing industry
I want to write in Python! (3) Utilize the mock
Ventilation is important. What I did to keep track of the C02 concentration in the room
How to give and what the constraints option in scipy.optimize.minimize is
What to do when you get "I can't see the site !!!!"
What to do when the value type is ambiguous in Python?
[Python] What to do when PEP8 is violated in the process of importing from the directory added to sys.path
What to do when the result downloaded via scrapy is in English
For the first time in Numpy, I will update it from time to time
What to do when the warning "The environment is in consistent ..." appears in the Anaconda environment
[At Coder] What I did to reach the green rank in Python
What is the fastest way to create a reverse dictionary in python?
What is "mahjong" in the Python library? ??
Why Docker is so popular. What is Docker in the first place? How to use
I want to make the LED Lighting of ErgoDox EZ shine, but tell me what the LED is in the first place