••• 8 ••• Innovationen Listening to the Laser Real-time Process Insight expanding the approach to laser welding and cutting applications. Quality assurance in laser micromachining often involves delays, technical effort, and complex post-processing. Typical structures feature micro- and submicron dimensions, requiring time-consuming microscopic inspection. Fraunhofer IWS now provides an industry-oriented alternative: a new acoustic monitoring module analyzes sound emissions generated during processing, assesses quality in real time, and visualizes anomalies – without removing the part from the machine. Acoustics as a Data Source: Real-time Evaluation during the Process The system uses robust semiconductor microphones to capture sound emissions, a local evaluation unit, and a neural network trained on real-quality data. It breaks the signals into frequency components and extracts key parameters such as amplitude, duration, and pulse shape. These flow into an AI-based analysis that reliably classifies the process state as stable, deviating, or faulty. Feedback occurs instantly and independently of additional sensors. A standout feature is “acoustic imaging”. The system generates a two-dimensional map of the structured surface based on the spatial sound emissions distribution. This acoustic image mirrors conventional surface inspection–without external metrology, purely from process data. Irregularities such as contamination, misalignment, or uneven ablation appear clearly and allow documentation. “We wanted a solution that evaluates process quality right at the machine – without sample extraction or lab measurement,” says Tobias Steege, Group Manager for Systems Engineering and Sensing at Fraunhofer IWS. “Our acoustic analysis does not replace microscopic inspection but enables robust, economical inline control for every single part.” Compact Integration, Cost-efficient Operation The monitoring module comes as a plug-and-play solution for industrial use. It features standardized Ethernet interfaces (TCP/IP, REST), needs no external cloud, and connects to standard machine controls. A modular design and use of semiconductor components significantly reduce system costs compared to conventional acoustic monitoring. This makes precise, inline-capable process control viable even for small production volumes. “The system runs out of the box – no special hardware, no external data link, just a network connection to the machine,” Steege explains. “While acoustic analysis cannot replace microscopic surface inspection, it ensures robust and economical quality control for each individual component.” Rooted in Research, Ready for Industry Several publicly funded projects laid the foundation for this development, including the EU-funded SYNTECS project and the German BMBF-funded initiative Medius, which explores data-driven methods for laser process monitoring. Fraunhofer IWS implemented a fully functional demonstrator through internal funding from the Fraunhofer Performance Center Smart Production and Materials, with a clear focus on industrial application. The institute offers industrial partners the opportunity to test the system in pilot applications or integrate it as an OEM component. “We see great potential for acoustic monitoring across industries – from high-precision structuring to robust welding processes,” Steege emphasizes. “Our strength lies in the combination of process-oriented sensor technology and intelligent software design geared at everyday industrial use.” Next Step: Monitoring for Welding and Cutting In addition to laser-based surface structuring, another team at Fraunhofer IWS, led by Dr. Axel Jahn, is transferring the principle of acoustic analysis to macro processing applications such as laser welding and cutting. These processes generate complex acoustic and optical emissions that directly reflect the quality of weld seams and cut edges. The researchers capture data using high-dynamic microphones and high-speed cameras to monitor these processes. They first separate proper signals from background noise and link them to characteristic process states and quality indicators. In collaboration with the Fraunhofer Center Mid-Atlantic (CMA) in the United States, Fraunhofer IWS is developing a dedicated sensor platform and tailored data analysis algorithms. The system uses high-dynamic microphones and optical sensors that integrate seamlessly into industrial welding cells. Machine learning algorithms analyze the sensor data and reliably detect key patterns such as beam misalignment, spatter formation, or seam instability. “Welding and cutting processes also produce acoustic signatures that reveal quality and highlight faults,” explains Jahn. “We design systems that capture these signals reliably – even in noisy environments. In the future, we aim to bring laser applications with sensor-based process monitoring into broader industrial use.” Fraunhofer IWS develops acoustic monitoring solutions that evaluate laser processes in real time –enabling robust, cost-efficient process control that detects deviations instantly Foto: Fraunhofer IWS Humanoid robots A huge growth market Tesla considers itself a technology leader with their humanoid Optimus robot. It moves elegantly at great speed and features a five-fingered hand of great dexterity. Elon Musk believes that they will soon be suitable for housework or childcare, and Goldman Sachs also sees great potential in the global humanoid market. They expect a market valuation of 38 billion USD by 2035. Apart from US companies, the People’s Republic is pursuing ambitious goals as well. The country intends to become the global market leader in the field of humanoid robots within three years. In the medium term, five percent of all jobs in China are scheduled to be taken over by humanoids. This is equivalent to the deployment of around 35 million units. Chinese companies across various applications and industries already deploy humanoids by Unitree, AgiBot, Engine AI, Fourier, and Ubtech. Industry leader Unitree is one of the world’s first to place a small humanoid on the market at a spectacularly low price-the G1. The basic version of the G1, standing 1.3 meters tall and weighing just 35 kg, will cost only 16.000 Euros in China. In Europe, however, the price is expected to be closer to 30.000 Euros. And this price can easily more than double depending on the robot’s configuration. The G1 has raised great interest, as Unitree Marketing Manager Qian Yuqi tells us: “We already have orders from China and abroad. Initially, our robot will probably be most popular among scientific research institutions, but we also know of industrial companies considering the G1 for simple tasks.” This shows: The time of humanoids has come and, thus, a new era of automation has begun. AI already empowers them to take on some tasks that could only be handled by humans thus far. automatica promises exciting insights into the current capabilities of humanoids and their potential impact on work environments. Continued from Page 1
RkJQdWJsaXNoZXIy NjM5MzU=