Honest Intelligence

Creating full transparency with customers, a TRUMPF webinar outlines the company’s current use and development of AI as well as its future strategies


Since its emergence, artificial intelligence (AI) has been changing the world – initially in ways that society might not have been fully aware of and, after the launch of ChatGPT, in ways that sparked international debate and scrutiny.

As with most things, though, AI is becoming better understood and appreciated as time passes. In fact, its use in manufacturing serves as a prime example of the good it can do. Bringing about increased productivity and competitiveness, AI can “evaluate machine data efficiently, detect errors and automate production,” explained Catherine Flynn, a public relations and communications specialist for TRUMPF, as she kicked off a webinar on the topic of AI on April 16, 2024.

Stephan Mayer, CEO, Machine Tools, TRUMPF SE + Co. KG, took the virtual stage at a recent TRUMPF webinar focused on artificial intelligence.

To explain how the company is positioning itself related to AI and the opportunities it creates, Stephan Mayer, CEO, Machine Tools, TRUMPF SE + Co. KG, took the baton from Flynn. During the hour-long webinar, he used the platform to discuss the investments the company has made to evaluate the potential of AI and the ways TRUMPF has injected AI into company products and internal processes – all for the ultimate benefit of customers.

In terms of product development, Mayer said that AI has presented a huge sea change. In the past – especially in the automotive and machine tools industries – typical software product lifecycles were five to 10 years. “If you have an idea for a new product, you bring the engineers together, design it and test it for half a year,” he explained. “Then, you have a test customer followed by a market launch where it lives for up to six years or so. And then, you replace it with a new product.”

But today, companies like TRUMPF are seeing faster development times – especially in terms of the launch of digital products. If TRUMPF maintained its historical approach of lifecycle times ranging from five to 10 years, “we would be far too slow,” Mayer said. “So, we had to change our R&D and product development process to a much more agile version that allows us to further develop and enhance existing software products on a three-month cycle.”

AI at work

Based on AI’s ability to recognize repeating structures and patterns within big data sets, Mayer stressed its use in industry and at TRUMPF for making statistical decisions and creating statistical outcomes. “Whenever you have repeating patterns, actions or processes, AI should be of interest,” he said. It can be a powerful tool to automate processes or speed them up.

To date, TRUMPF has already applied AI to multiple areas that are of interest to customers, including preventive maintenance and quality control. With AI at work, a laser can be constantly monitored, watching for specific cutting parameters or even the temperature of the laser light’s reflections. It can also be employed during welding operations, carefully watching the quality of the seam as the bead is laid down. If undesirable conditions are identified, customers can correct an issue before it occurs or before it rises to the level of scrapped material, downtime or a service call.

TRUMPF’s AI solution for identifying spare parts is a good example of the strides the company is taking to leverage AI in customer offerings. Thanks to the technology, customers can automatically place a spare parts order by uploading a cellphone photo into the company’s app.

Service is another area where TRUMPF is leveraging AI. When company technicians make a service call of any kind, they write a service report, which is documented and then stored in a larger historical database.

“We have hundreds of thousands of these reports and with a large language model, similar to ChatGPT, we can apply this big database of information for first steps in correcting issues,” Mayer explained. “When customers contact us to solve an issue, the large language model can research the database and provide the right answer. We are already applying this in the service department to speed up the process of resolving issues.”

Intelligent advice

Considering AI’s ability to recognize patterns, it can and should be utilized to help customers running multiple shifts. Based on the data that TRUMPF has accumulated, AI can help determine which parts are better to run during the day when people are in the shop versus at night when machines are running unattended.

“We have collected hundreds of thousands and perhaps even millions of different shapes that have been successful or unsuccessful in the process of automatic loading, unloading and sorting,” Mayer said. “So now, we have trained this algorithm that can tell [customers], for example, this is a complex shape, which means there is a high probability that automatic loading, unloading and sorting might not work.”

To illustrate this capability, Mayer offered the example of automatically unloading cut pieces without getting them stuck within the sheet metal skeleton. To make these determinations, successful and unsuccessful parts are logged into a database for the AI to consider.

Because of AI’s ability to recognize repeating structures and patterns within TRUMPF’s massive service report database, the process to resolve issues has been significantly sped up.

“Every time a machine was successful in unloading a part, the machine realizes, ‘this shape was successful,’” he explained. “And what we do now is give this information back to the customers. When a customer is programming a machine and puts all the different parts on the sheet, the machine can tell the operator, ‘Hey, be careful; there is a part on that sheet that is complicated, so you should process this during the day shift when an operator is there to fix any problems that might arise instead of processing it during the night shift when the machine might stop.’”

It’s a scenario that explains the growing power of AI in the world of sheet metal, Mayer added. In that regard, AI is constantly evolving and improving over time. In fact, TRUMPF’s AI has matured to a level where it will not only retry to remove a difficult part from a skeleton, but it will also make recommendations for where to place a suction cup or grabber in the future for better part removal outcomes.

Customers can also rely on AI for real-time cutting advice. Take the common example of processing material that is rusty or has varying thickness across the sheet.

“We place a high-speed camera on top of a laser that is pointed at and interpreting the melting spot of the laser while it’s cutting,” he explained. “With the help of AI, we interpret the picture and find out when there is an issue with the melting spot, which could dictate that the machine automatically lowers its speed, for example.

“This also works in reverse,” he added. “If everything goes right, the machine can accelerate. This is what we call Active Speed Control. It’s like an autonomous driving system.”

Assembling the experts

Throughout TRUMPF’s years-long push to bring AI to its customers, the company has assembled a team of more than 100 people involved in AI applications. Along the way, leadership identified a shift in its hiring practices. There will always be a need for typical machine engineers and mechanical engineers, but the need for software and AI developers is clearly growing.

“We need experts to understand what AI can do and what AI cannot do,” Mayer said. “Typical engineers don’t have experience in AI, so we’ve hired AI experts from universities and from other sources and have assembled teams that consist of sheet metal application engineers and AI experts. They are constantly talking to each other to try to figure out the problems that AI can actually solve.”

TRUMPF has trained an AI model to help customers determine which parts are better to run during the day when people are in the shop versus at night when machines are running unattended.

No matter the combined experience of the assembled teams, Mayer stressed the importance of data in the AI equation. TRUMPF needs data because AI needs data, he said. Some of the data at TRUMPF was pre-existing in the company’s internal databases, but when it comes to real-time data processing and supporting customers in their processes, their data is critical, too. Connected machines will be required to collect further data for the AI to learn and mature.

“We are at the beginning of a long journey, but we see that the development of AI’s capability is happening fast,” he said. “We at TRUMPF are in the early stages, and … every day, we see something new.

“It will take time,” he concluded, “but [AI] is a promising improvement for automation in its journey. We see the lack of skilled labor is continuing, so AI can help us reach the next S-curve.”


Get industry news first
Subscribe to our magazines
Your favorite
under one roof