DFG Research Unit KI-FOR 5339

AI for the fast maturation of new manufacturing processes

Thermoplastic composites promise light parts that can be formed automatically in short cycles and recycled. A new material or part design, however, can make an established process immature again, which leads to costly trial and error. We develop AI methods that learn from every trial and carry this knowledge over to new variants, so that they mature with fewer trials.

Karlsruhe Institute of Technology and Fraunhofer IOSB
Funded by the Deutsche Forschungsgemeinschaft since 2023

Illustration of our aim. Every new geometry or material starts as an immature process, and knowledge carried over from earlier variants lets it mature in fewer trials.

Why new processes are hard to get right

Four factors make new processes slow to mature. Today, experts bring them to maturity through extensive experiments guided by their experience.

Demanding materials

Fibre-reinforced thermoplastics are stiff along the fibres and soft across them. Stacked in layers, they deform in ways that depend on temperature and speed.

Many coupled settings

Temperatures, speeds, forces and gripper positions interact, and tuning the conventional settings alone does not prevent wrinkles.

Costly, late feedback

Defects such as wrinkles show up only after forming. Every trial costs machine time and material, and some parts must be destroyed to test them.

Every change starts over

A new material or geometry usually means repeating the experiments. What was learned on one variant rarely carries over to the next.

How the subprojects work together

T1 provides the demonstrator process and its experiments. T2 simulates the process, and M1 decides which sensors observe it. M2 learns fast process models from these data, mainly from T2's simulations. M3 optimises the process settings, and M4 estimates the process state and controls it. Across all stages, F defines measures of maturity and records the process data in a knowledge graph.

The example process

The Research Unit tests its methods on non-isothermal stamp forming of thermoplastic tape laminates, an immature process with a high potential for improvement. The process runs on a demonstrator in the Karlsruhe Research Factory and in the stamp-forming laboratory of Fraunhofer ICT.

Illustration of the demonstrator line. A carriage moves the laminate in its gripper frame to the infrared oven, the press and optical inspection. In the press, the grippers follow the laminate as it is drawn into the die. Colour shows the laminate temperature.
The stamp-forming line with gripper frame, infrared oven and press on a linear axis, with an industrial robot in the background.
The demonstrator in the Karlsruhe Research Factory.

The material

The published forming experiments used unidirectional carbon-fibre tapes in a polyamide 6 matrix, stacked into laminates with different fibre directions (Zeeb et al. 2025).

The process

An infrared oven melts the matrix. The press then forms the hot laminate, which solidifies as it cools in the tool. Wrinkles are the main defect we study.

Highlighted publications

Ten of our 72 publications, chosen to cover every subproject. Members of the Research Unit are shown in dark type.

  1. Composites Part A 2026

    Bayesian optimization and guided over-instrumented gripper design for thermoforming of composite materials

    Frank Doehner, Johannes Mitsch, Saksham Kiroriwal, Shahenda Youssef, Georg Zeeb, Frank Henning, Luise Kärger, Jürgen Beyerer

    T1 T2 M1 M3

  2. Composites Part A 2026

    Rate-dependent 3D forming simulation of thermoplastic composite materials using visco-hyperelastic material modeling and 3D hexahedral solid-shell elements

    Johannes Mitsch, Bastian Schäfer, Luise Kärger

    T2

  3. CMAME 2024

    Physics-informed MeshGraphNets (PI-MGNs): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes

    Tobias Würth, Niklas Freymuth, Clemens Zimmerling, Gerhard Neumann, Luise Kärger

    T2 M2

  4. NeurIPS 2025 Spotlight

    Diffusion-Based Hierarchical Graph Neural Networks for Simulating Nonlinear Solid Mechanics

    Tobias Würth, Niklas Freymuth, Gerhard Neumann, Luise Kärger

    T2 M2

  5. NeurIPS 2025

    AMBER: Adaptive Mesh Generation by Iterative Mesh Resolution Prediction

    Niklas Freymuth, Tobias Würth, Nicolas Schreiber, Balázs Gyenes, Andreas Boltres, Johannes Mitsch, Aleksandar Taranovic, Tai Hoang, Philipp Dahlinger, Philipp Becker, Luise Kärger, Gerhard Neumann

    T2 M2

  6. NeurIPS 2025

    MaNGO – Adaptable Graph Network Simulators via Meta-Learning

    Philipp Dahlinger, Tai Hoang, Denis Blessing, Niklas Freymuth, Gerhard Neumann

    M2

  7. ICLR 2025 Oral

    Geometry-aware RL for Manipulation of Varying Shapes and Deformable Objects

    Tai Hoang, Huy Le, Philipp Becker, Ngo Anh Vien, Gerhard Neumann

    M2 M4

  8. IFAC World Congress 2026

    Smooth Sampling-Based Model Predictive Control Using Deterministic Samples

    Markus Walker, Marcel Reith-Braun, Tai Hoang, Gerhard Neumann, Uwe D. Hanebeck

    M4

  9. IEEE INDIN 2024 Best Student Paper Award, AI for Industry track

    Joint parameter and state-space modelling of manufacturing processes using Gaussian processes

    Saksham Kiroriwal, Julius Pfrommer, Hendrik Mende, Robert H. Schmitt, Jürgen Beyerer

    M3

  10. IEEE CASE 2026

    LLM-Based Pre-Press Decision Support for an Industrial Composite-Forming Workflow

    Yuyang Li, Alexander Bott, Jürgen Fleischer, Tobias Käfer

    T1 F

All publications

News

  1. Publication

    Three papers accepted at NeurIPS 2026

    Two of the papers come from subproject M2, one of them developed with T2, and one comes from subproject M3.

  2. Publication

    Two articles in Composites Part A: Applied Science and Manufacturing

    An article from subproject T2 appeared in January, and one from subprojects M1, M3, T1 and T2 followed in August.

  3. Publication

    Paper at ICLR 2026

    The paper “Improving Long-Range Interactions in Graph Neural Simulators via Hamiltonian Dynamics” from subproject M2 appeared at ICLR 2026.

All news