Industry experts discuss how machines and people can enhance manufacturing outcomes.
A highly intelligent, interactive and automated manufacturing ecosystem fueled by advanced artificial intelligence (AI) is on the way, but a skilled workforce is still needed to provide the necessary human touch.
Caterpillar, Inc. Digital Operations and ERP Strategy Team Leader Brent Ruth spoke at a July AI in Manufacturing—Cognitive Manufacturing 2021 conference in Germany about what’s underway during the current Industrial Revolution 4.0 (Industry 4.0), as traditional analog manufacturing has maxed out its efficiency limits and human workers are reintegrating into the digital age.
“It’s now possible to connect individual machines and assets into a globally connected network that not only provides new data and insights, but learns and adapts to changing requirements, the network can sense, and predict actions and outcomes, and drive improvements,” Ruth said.
The exponential nature of technological growth is captured by Moore’s Law, which states that the number of transistors in dense integrated circuits doubles about every two years and serves as the backdrop for the current digital revolution.
“The iPhone 6 is 32,600 times faster than the best Apollo era computers that put a man on the moon and could perform instructions 120 million times faster, and the iPhone 11 is twice as powerful as the iPhone 6,” Ruth noted.
Running parallel to the Industry 4.0 is the emergence of a Workforce 4.0, which according to Ruth, isn’t about robots replacing humans but rather about taking the robot out of the human. The duties of many jobs are repetitive, menial and mundane, and they don’t inspire curiosity or creativity. The role of technology can be to take over these roles, allowing people to pursue higher value-added activities. The catch is that this path involves a growth mindset where the existing workforce is retrained and reeducated for future jobs.
AI in the manufacturing sphere can be a crucial factor for a successful business with an up-skilled workforce equipped to utilize the new technologies. For example, global automotive technology company Faurecia is currently reinventing its approach to achieving industrial processing excellence by launching a combined collaborative and augmented intelligence initiative to boost knowledge and productivity in 33 of its 53 factories. As a result, one of the plants managed to reduce the amount of scrap produced as a by-product of the manufacturing process by 10 percent in two months by using AI.
On the factory floor, the integrity of real-time data is paramount, and Faurecia uses algorithms to ensure that the data is of good quality, harnessing the speed of modern computing power to display the data as quickly as needed.
According to Faurecia Injection Senior Manager Stephane Coudurier Curveur, the company has set up digital connections between factory machines and operators, who monitor progress via screens. A digital record of everything that occurs during the production process is recorded in real time in a way that would be impossible for humans to undertake without the technology. Whereas in the past, production issues that arose would require time-consuming investigations, the information is now digitally transmitted, and operators can analyze the data and respond to it. AI then makes future predictions as well as comparative predictive models to serve as a guide for operators.
For example, operators at one factory noticed that there was a slight increase in scrap production. Data analysis of the good batches and the ones that yielded an increase in scrap was then conducted. Experts could then analyze the situation, and the algorithm pointed to a particular sensor that indicated a temperature variation with the batches. The scenario resulted in a 4 percent production loss, which Curveur said could be avoided by employing predictive analysis.
In another example, the factory operators frequently changed the type of product made in small batches and saw a high degree of variation in production. The data science indicated that batches in a particular zone were the cause of the issue. A maintenance crew was deployed to the machine, where they discovered that the wrong sensor had been replaced during a previous maintenance session. They also found that a leak was causing air to enter the machine’s hydraulic oil, causing further production problems.
“So, once we had investigated and found the remedy and corrected, the machine was back to normal,” said Curveur. “Before the work, it was taking 15 minutes before stability. And after the work, it was 10 minutes to stabilize, which is the normal time. And here, just by luck as with the other case, it’s another 4 percent we saved.”
Curveur said that the goal is to have people act on the data, which the AI verifies. This results in time savings and quicker response times. In addition, the tech tools allow for greater perspective and performance analysis, such as exposing the best machines, the worst machines, the margin by which production targets were missed, and how much scrap was produced.
“So, with our data scientist, we defined three types of algorithms that we believe can help us in 80 percent of the cases, then people have access to the decision tree, and they can target the machine and the tool, and they will have analysis, and analysis of which parameter was driving most of their problems.”
One of the findings made by Faurecia managers is that the notion that only data scientists can work with AI systems is a fallacy. The company offered simple AI tools to factory workers who were able to effectively employ them for better performance.
“We think we can solve 80 percent of our problems with basic algorithms, and for the most complicated ones, we have our own data scientists to deal with it,” said Curveur. “And you can focus more on continuous improvements, which means this is where you really make money—dealing with everyday issues is not where you make improvements; you improve when you make continuous improvement.”
Most manufacturing companies have yet to adopt AI for operational efficiency, and some reportedly have not yet experienced a return on investment. Part of the issue is that AI manufacturing is still in its infancy, and there’s a learning curve for workers. However, AI is gaining ground in many industries.
According to a report from the Capgemini Research Institute, AI in manufacturing is a “game-changer,” and Europe is leading the way with a 51 percent adoption rate among the continent’s manufacturers, followed by 30 percent in Japan, 28 percent in the U.S., 25 percent in Korea, and 11 percent in China. The three main AI use cases are for intelligent maintenance of plant machinery and equipment, product quality inspection, and demand planning. However, the largest share of AI uses—29 percent—are for maintenance.
The report concluded that while there’s potential for an AI revolution in manufacturing, it’s still mainly in the experimental phase with few implementation examples at scale.
“The challenge that we have is that it’s not only a question of technology,” said Luis Miguel del Say Rodriguez, head of Digital, Design, Manufacturing Services for Airbus Commercial Spain. “It is also a question of people and cultural change and how we are going to deal with it. It’s clear that with all these AI tools, we have significantly increased the level of information. But we need to convince the end-user of the insights regarding the dependability of the data generated using AI.”