Fraunhofer Institute for Manufacturing Engineering and Automation

Künstliche Intelligenz kommt immer mehr in Produktionsumgebungen an. Hier kann sie genutzt werden, um Maschinenparameter automatisiert und basierend auf Sensordaten aus der Produktion oder anhand von Bilddaten eines Bauteils zu optimieren. Am Fraunhofer IPA sind verschiedene KI-gestützte Technologien entstanden, die sowohl die Bewertung des Ist-Zustands eines Bauteils als auch daraus resultierende Verbesserungen der Produktionsparameter ermöglichen.

Background The World Health Organization Emergency Care Systems Framework (WHO ECSF) was designed to offer guidance in establishing and developing effective Emergency Medical Services (EMS) systems. However, evolving disease patterns, changing community needs, and a rising demand for emergency care services, highlight the need for more integrated and patient-centered EMS systems. This evolution should be mirrored in the WHO ECSF. Hence, this study explores system components of the Copenhagen (CPH) EMS that may enhance the WHO ECSF´s emphasis on integrated and patient-centered care. Methods A qualitative case study was conducted from April through June 2021, including (i) semi-structured interviews with researchers and professionals at the CPH EMS and (ii) a scoping literature review using PubMed, Google Scholar, expert recommendations and snowballing. Results Thirteen expert interviews and 35 records were analyzed, revealing key integrated care components within the CPH EMS. These include education and citizen participation programs, early triaging, differentiated care pathways coordinated with primary care and out-of-hours services, and specialized mobile care units complementing “traditional” ambulance services. Technology supports integrated and patient-centered care by facilitating early differentiation of care, efficient dispatching, and communication. Data-driven approaches were fostered through technology-aided data collection, supporting research, quality improvement, and patient safety. The identified components were mapped within the WHO ECSF´s four domains: scene, transport, facility, and cross-cutting elements. Due to the prehospital focus of the CPH EMS, limited data was available for the “facility” site. Conclusions The CPH EMS demonstrates an integrated, patient-centered systems approach that emphasizes seamless coordination along the patient care pathway, bridging EMS with broader health and social systems. Research-informed initiatives and intelligent technology solutions underscore the potential for enhancing the WHO ECSF. These findings highlight the importance of continued system integration and a holistic health perspective, including in emergency settings. Further research is needed to assess the transferability of these components across diverse global contexts. Trial registration Not applicable.

Due to a combination of high surface quality, rigidity, and low tool change times, shrink fit tool holders are a popular choice for machining tool clamping. However, tool slippage and runout can be observed at high-performance machining, especially with low tool diameters. This study provides a possible solution to prevent slippage in shrink fit tool holders. For this, honeycomb-like laser structures are applied to circumferential surfaces of fine-grained 4-mm carbide tool rods with different material compositions. Laser structuring is carried out with a Yb:YAG infrared laser with a pulse duration of 900 fs. In preliminary tests, the ablation behavior of the laser is studied for different materials. Using this data, laser structures with different structure depths and spot-to-spot spacings are generated by varying the laser parameters. Laser-structured tool rods are then clamped in a specifically designed test bench using a radial clamping element and loaded until slippage occurs. Laser-structured tool rods achieve a slip load up to 2.4 times higher than their unstructured counterparts. Slippage almost exclusively occurs due to the wear of the radial clamping element used in the test bench. Lower spot-to-spot spacings as well as higher structure depths generally result in higher slip loads. However, at structure depths of 15 µm or greater, tool rods fracture, which is attributed to the notch effect. An effect of the composition of the tool rods on the slip loads or laser ablation behavior could not be observed.

Quantum machine learning (QML) models use encoding circuits to map data into a quantum Hilbert space. While it is well known that the architecture of these circuits significantly influences core properties of the resulting model, they are often chosen heuristically. In this work, we present a approach using reinforcement learning techniques to generate problem-specific encoding circuits to improve the performance of QML models. By specifically using a model-based reinforcement learning algorithm, we reduce the number of necessary circuit evaluations during the search, providing a sample-efficient framework. In contrast to previous search algorithms, our method uses a layered circuit structure that significantly reduces the search space. Additionally, our approach can account for multiple objectives such as solution quality and circuit depth. We benchmark our tailored circuits against various reference models, including models with problem-agnostic circuits and classical models. Our results highlight the effectiveness of problem-specific encoding circuits in enhancing QML model performance.

Additive manufacturing (AM) offers design freedom and cost‐effective production of complex parts. Powder bed fusion of polymers using laser beam (PBF‐LB/P) is one of the most industrially relevant plastic AM techniques. Although the industrial importance of PBF‐LB/P is steadily increasing, the range of materials available is very limited. This study combines material and process data to determine the energy conversion using dimensionless parameters for unmodified and nanoadditive‐modified polyamide 12 (PA12) powders and polypropylene (PP) powder. Monolayers with different energy inputs are printed and their thickness is measured. The process is described by a dimensionless energy input value and an energy demand value to characterize the energy conversion during processing, considering important material and process parameters. Tensile tests are performed to verify the finding of suitable printing parameters. Furthermore, multilayer experiments are carried out to replicate the printing process in an application‐oriented experiment. This study investigates the influence of the interlayer time (ILT) on the resulting part properties of PA12 and PP. Experiments and numerical models are developed to study the stepwise thickness increase of the first ten layers and density of multilayer samples. The results demonstrate increasing density with higher ILTs, which is in accordance with the simulation predictions.

Zusammenfassung Hintergrund In medizinischen Registern werden wertvolle Daten von Patienten gesammelt, die die Qualität und Wirksamkeit von Behandlungen überprüfen bzw. kontrollieren. Es existieren nationale Register im Bereich der Amputationsmedizin und Patientenversorgung wie das Swedish Amputation and Prosthetics Registry (SwedeAmp) und das Limb Loss and Preservation Registry (LLPR) in den USA, die Informationen zu Prothesenarten, -materialien und -verfahren sowie patientenbezogene Ergebnisse wie Mobilität und Lebensqualität erheben. Seit 2011 konnte SwedeAmp wichtige Erkenntnisse zu Langzeitergebnissen nach Amputationen beitragen und die Versorgung in Schweden verbessern. Das LLPR in den USA erfasst Daten von klinischen bis zu psychosozialen Aspekten, was länderübergreifende Vergleiche und die Optimierung der Versorgung ermöglicht. Material und Methoden In Deutschland leistet das AMP-Register bedeutende Beiträge, indem es u. a. Daten zu Prothesenpassform und -tragekomfort sowie Gründe für Revisionen dokumentiert. Ziel ist es, eine Evidenzbasis durch systematische Datenerhebung zu schaffen. Das Projekt umfasst den Aufbau einer benutzerfreundlichen IT-Struktur, eine Pilotphase zur Anwendungsevaluierung und die enge Zusammenarbeit mit Experten. Mithilfe standardisierter Datensätze sollen Versorgungsdefizite aufgedeckt und evidenzbasierte Ansätze entwickelt werden. Datenerfassung und -speicherung erfolgen gemäß der Datenschutz-Grundverordnung (DSGVO) und werden durch technische Maßnahmen abgesichert. Ergebnisse und Diskussion Erste Ergebnisse am Studienzentrum Heidelberg zeigen das Potenzial des AMP-Registers. Subgruppenanalysen unterstützen die Versorgungsoptimierung und bestätigen die Relevanz regelmäßiger Assessments, um die Versorgungsqualität langfristig zu verbessern.

Anomaly detection is becoming increasingly important and has found its way into manufacturing applications. The potential is seen in use cases such as maintenance cost reduction, machine fault reduction, or increased overall production based on industrial time series data. However, obstacles arise in practice. Supervised algorithms lack limited and expensive labeled training data, and unsupervised algorithms do not have the capabilities for evaluation and tracking. We propose a data-efficient architecture for anomaly detection using energy consumption time series data to address these limitations. To do so, we design an active learning model that optimizes an unsupervised model by integrating budgeted expert feedback. Our solution builds on an autoencoder to leverage latent space representations for an additional supervised feedforward network trained with expert knowledge labels to distinguish between normal data and anomalies. Four different strategies for querying the still-unlabeled data are compared so that the expert’s resources are used efficiently. We validate our concept in an industrial robotic screwdriving application based on energy data for condition monitoring. Findings for the application tested indicate that anomaly detection performance can be significantly increased by 59% for the F1 score with active learning compared to unsupervised models. Furthermore, models trained only on energy consumption data exhibit the same performance as models trained on difficult-to-obtain mechanical process data, thus confirming the practicality of our proposed approach and data efficiency for the use of easily accessible energy data in manufacturing applications. While our approach enables an active learning model to be added to an existing unsupervised model, it allows for straightforward benchmarking and extension to other manufacturing applications.

Recent advances in piezo inkjet printing technology have enabled the use of high-viscosity materials previously unsuitable for inkjet applications. This technology facilitates the printing of vat photopolymerization resins, traditionally used to produce single-material components with superior properties compared to those achieved with conventional UV inks in material jetting. The ability to print these resins with piezo inkjet technology opens new fields of application for producing multimaterial functional parts. To fully exploit this potential, it is essential to evaluate the compatibility of these resins with the new technology. This study systematically evaluates the jetability, printability, and performance of high-viscosity materials to optimize the printing process and provides a detailed understanding of the factors influencing the jetability of high-viscosity resins and develops guidelines for optimizing their use in 3D inkjet printing applications.

sQUlearn introduces a user-friendly, NISQ-ready Python library for quantum machine learning (QML), designed for seamless integration with classical machine learning tools like scikit-learn. The library’s dual-layer architecture serves both QML researchers and practitioners, enabling efficient prototyping, experimentation, and pipelining. sQUlearn provides a comprehensive tool set that includes both quantum kernel methods and quantum neural networks, along with features like customizable data encoding strategies, automated execution handling, and specialized regularization techniques. By focusing on NISQ-compatibility and end-to-end automation, sQUlearn aims to bridge the gap between current QC capabilities and practical machine learning applications. The library provides substantial flexibility, enabling quick transitions between the underlying quantum frameworks Qiskit and PennyLane, as well as between simulation and running on physical hardware.

High Speed Sintering (HSS) is an additive manufacturing process with great potential to produce complex, high-quality polymer parts on an industrial scale. However, little information is currently available on the characteristics of the powder materials used and the part properties that can be achieved. This is also the case for the standard material polyamide 12 (PA 12) and the first commercially available HSS machine, the VX200 HSS. The aim of this study is, therefore, to provide the first comprehensive overview of the properties of PA 12 parts manufactured with the VX200 HSS and the characteristics of the PA 12 powder. This includes the analysis of the influence of part orientation and part position in the build job. To characterize the powder, particle size distribution, particle shape, thermal properties and powder flowability were analyzed. To characterize the parts, density and various thermal, mechanical and geometrical properties were analyzed. In summary, the powder material and part properties are largely similar to those of other Powder Bed Fusion of Polymer (PBF/P) processes. However, there are also significant differences for some part properties. It was also determined that the anisotropy of parts is very low compared to that of many other additive manufacturing processes.

We have been intending to realize a quadruped robot controller autonomously generating rhythm and gait while utilizing natural body dynamics under the gravity. For this purpose, we newly proposed a model of the rhythm generator mainly consisting of sensorimotor functions in the spinal cord, and simulated the walking–running transition of a low spinal cat with hindlimbs according to increased belt speed on a treadmill (Forssberg et al. 1980). In this simulation, we constructed a physical model of the hind-legged biped robot with a trunk and fixtures, and used an independent leg controller with such rhythm generator for each leg. When we employed hip flexion/extension and leg unloading as sensor information for the stance–to–swing phase transition of the rhythm generator, the rhythm in the steady walking could be generated mainly based on leg unloading. In the transient state under the perturbation of belt speed change and the steady running, different rhythms could be autonomously generated based on the combination of leg unloading and hip flexion/extension. Moreover, the walking–running transition could emerge by the change of the dynamics structure triggered by the increase of belt speed.

The location of defect formed in the final composite is identified using sensor data. Herein, we report the development of an online process monitoring system for vacuum‐assisted resin transfer molding (VARTM) process using large area graphene coated in‐situ fabric sensor. Besides imparting excellent mechanical properties to the final composites, these sensors provide critical information during the composite processing including detecting defects and evaluating processing parameters. The obtained information can be used to create a digital passport of the manufacturing phase to develop a cost‐effective production technique and fabricate high‐quality composites. The fabric sensor was produced using a scalable dip‐coating process by coating 1‐, 3‐ or 5‐layers of thermally reduced graphene oxide (rGO) onto glass fabric surface according to the number of dips of the fabrics into GO solution. 5 electrode pairs were placed in the horizontal and vertical directions on the area of each coated fabric sensor before placing it inside the VARTM setup. The electrical resistances from all electrode pairs were simultaneously and continuously recorded during distinct stages of the VARTM process to determine the relative conductance. During the vacuum cycle, the range of relative conductance increased with the number of coated rGO layers, with the 5‐layer rGO‐coated sensor showing the highest conductance range of 16.9 %. Additionally, it was observed that the 5‐layer coated sensor showed a consistent decrease in conductance during the infusion phase due to the fluid flow pressure dominating the resin electrical conductivity. Most importantly, physical parameters such as infusion time, flow front location, race‐tracking and dry spots were monitored in‐situ.

Traumatic Brachial Plexus Injury (TBPI) causes arm paralysis, impacting daily activities and reintegration into work. This paper presents the results of the first round of a quantitative Delphi study. The aim is to gather insights from five expert groups. The investigated user requirements primarily encompass design aspects like weight, fitting, and adjustability to accommodate adaptations. Besides increased functionality, a benefit for psychological recovery is presumed.

Microwave ablation therapy is frequently used to treat liver malignancies. To ensure proper tumor treatment, intraoperative feedback regarding ablation performance and lesion size is required. By employing an electrode array around the ablation needle, changes in electrical impedance of ex vivo liver are measured in real time. Time-series trends of magnitude and phase are measured for 90 °C and 110 °C ablation temperatures. A finite element model is additionally configured to simulate the underlying biological processes. Gradients in the magnitude and phase trends can indicate the growth of the ablation zone. In combination with a preoperative simulation, impedance-based ablation monitoring can be a possible tool to improve future treatments.