The German manufacturing technology company Trumpf has set out its vision to become a leader in its sector in the application of Artificial Intelligence (AI) – in both its offering to customers and in its own operations.
Berthold Schmidt, Chief Technology Officer at Trumpf, said, "In five years' time, we want to be the leading user and leading provider of AI solutions in our industry. By then, there should no longer be a job at Trumpf that does not have some connection to AI."
The company will strengthen and coordinate its activities in the future with a new competence centre. The aim is to make internal processes even more efficient with the help of AI. The company sees potential for increasing productivity in, for example, software development and administrative areas.
As regards its customer offerings, Trumpf wants to bring more AI innovations to the market in the machine tool, laser technology and electronics sectors.
For example, its iAssist softwware helps designers improve part design to improve quality and reduce the workload on designers.
“With iAssist, designers can achieve very good results much quicker,” says Jörg Heusel, R&D manager for sheet metal design at TRUMPF. “Instead of improving the parts by means of trial and error, the software automatically shows where the greatest potential for optimisation lies. This generates better results and helps companies become more efficient.”
iAssist also takes economic factors into account and shows, for example, how material can be saved or production steps eliminated.
At present, TRUMPF is making iAssist available free of charge to all interested users. “Customers can use the software to optimize their components,” Heusel explains. “We, in turn, can then use their data to further train the AI and continuously improve the quality of its analysis.”
Another example is an AI-based run-in assistant that helps production employees run in the Trumpf’s TruMatic 5000 punch laser machine more quickly and more easily. At present, when using the machine to produce a new order with a new geometry or from new materials, users must individually test the machine program. This results in idle machine time that can quickly add up, especially with fully automated machines that produce many different parts.
Based on various evaluation models, the new run-in assistant shows for each new job whether the TruMatic 5000 can produce parts directly or whether manual intervention will be required.
For example a common problem when laser cutting parts is that the cut parts don’t fall out of the sheet and need to be manually removed – by programming a machine learning algorithm with data from thousands of part you can predict the ones that are likely to get stuck. You can then make sure that these are run during the daytime rather than on an unattended night shift.
“Our Runability Guide gives companies an advantage in productivity and competitiveness,” explains Jonathan Eberle, project leader at Trumpf’s Development department. “This not only saves them time but also means they can then use employees’ skills for other value-creating tasks or for training new personnel.”
Trumpf’s Runability Guide uses various models to determine a part’s complexity. For this purpose, development engineers continually upload knowledge from production experts into the cloud. In addition, the software uses physical simulations and AI in order to identify any possible problems in each of the process steps. Last but not least, the solution also works with genuine machine data from TruMatic 5000 users. This enables it to draw inferences about potential issues during processing. “Once all the models have been used to evaluate the job, the solution delivers its assessment,” Eberle explains. “Depending on the part, this would take up to an hour with classic physical simulation models. But by using AI, we can significantly accelerate this process, meaning that our customers can use it virtually in real time.”
TRUMPF is already working with AI to make its own processes more efficient. One example is the pilot project for an AI-based language model for technical service, which works in a similar way to Chat GPT. If a machine malfunctions, the service engineer can ask the AI how to solve the problem. As the software continuously learns from TRUMPF's service reports, the quality of the answers is also constantly improving. This reduces the workload on service staff and machine errors can be rectified more quickly.
Stephan Mayer, CEO Machine Tools at Trumpf gives another example: “When we do the final assembly on a laser machine we put a microphone in it. We then run the different axes – X,Y and Z – and we listen to the sound the machine is making as it is moving.
“The AI is listening to the sound and the feedback from the machine as to whether the axis is running smoothly or if there is something wrong, a vibration that could indicate that a screw has not been tightened enough, or a screw is missing. If a screw is missing the sound is different.
“The AI is so good at hearing that you only need a microphone that costs a few Euros and we get feedback on whether the final assembly is good. It was shocking to discover that every 100 machines something goes missing – no nothing goes missing.
“And now we are applying this also in the field. So if a customer calls and says, ‘something's wrong with my machine, we let the machine run again and we listen to it with the microphone – which we left on the machine when we shipped it – and AI can tell us if the axis has an issue because there is something wrong with the sound.