ML4CPS – Machine Learning for Cyber-Physical Systems - Professur für Informatik im Maschinenbau

Conference in Berlin, March 19-20, 2026

About the Conference

Cyber-physical systems are required to adapt to changing demands, often experience architectural changes over their lifetime, and generate a heterogeneous set of data. All of this leads to significant demands on monitoring and control software. This conference focuses on aspects of machine learning and related domains, such as predictive maintenance, self-optimization, fault diagnosis, re-planning, and reconfiguration. To build intelligent cyber-physical systems close cooperation between AI-research and industrial engineering is necessary. To facilitate such an exchange is the goal of this conference.

The 9th Machine Learning for Cyber Physical Systems (ML4CPS) conference offers researchers and users from various fields an exchange platform. The conference will take place March 2026, 19th till 20th at the Fraunhofer Forum in Berlin. Hosts are Fraunhofer IOSB, Helmut Schmidt University, Hamburg University of Technology, and the Chair of Production Engineering of E-Mobility Components (PEM) of RWTH Aachen.

In this year’s edition of the conference we also encourage submission of contributions from industry detailing scientific problems they encounter. To contribute, an extended abstract of a maximum of one page is required and must be submitted through the conference portal.

Papers may cover, but are not limited to the following topics:

LLM-Agents for CPS: Large multimodal models for text, images, and time-series data offer new opportunities for industrial applications. They can unlock novel opportunities for intelligent automation and the increase of the overall performance and functionality of cyber-physical systems.
Physics-Inspired ML: Prior knowledge can be integrated into the neural network, through the network architecture, additional data from simulations, or imposing constraints on the loss function. This can be crucial for building robust and reliable Neural Networks.
Industrial AI: Integrating AI into manufacturing processes can help to optimize them and enhance operational efficiency. Still, integrating AI into legacy systems and existing infrastructure is still a major challenge.
Green AI: Reducing the energy consumption of AI systems is essential for industrial and edge applications. This topic focuses on methods for energy-efficient models, and the trade-off between performance and resource usage.
Hybrid Methods & Hybrid Systems: Hybrid methods integrate multiple learning and modeling techniques while hybrid systems combine discrete and continuous dynamics and, thus, are powerful paradigms for complex CPS and industrial processes. Methods related to data-driven model identification, diagnosis, verification, and analysis are relevant challenges for the community.

Agenda

Day 1: 19.03.2025

09:00 – 09:15:

Welcome Coffee

09:15 – 10:00: Keynote: Mingabyte

10:00 – 10:30: Coffee Break

10:30 – 12:00: Session 1: ML Methods & Adaptive Learning

10:30 – 11:00: U-Test Certainty Adaptive Tree for Regression on Non-Stationary Data Streams – Benedikt Stratmann (Fraunhofer IOSB)
11:00 – 11:30: State-Learning of Time Series Data with Contrastive Learning – Phillip Johann Overlöper (Helmut-Schmidt-Universität)
11:30 – 12:00: Towards Adaptive Traffic Pattern Clustering Reinforcement Learning in Traffic Signal Control – Magnus Redeker (Fraunhofer IOSB-INA)

12:00 – 13:00: Lunch break

13:00 – 14:30: Session 2: Manufacturing

13:00 – 13:30: Enabling Future Inline Quality Management: Early Battery Cell Quality Forecasting Using Machine Learning with Electrochemical Impedance Spectroscopy and Incremental Capacity Analysis – Rui Yan Li (RWTH Aachen)
13:30 – 14:00: Development of a Quality Monitoring System for an Adaptive Vacuum Laser Welding Process for Hairpin Stators via Deep Learning – Yazan Bajah (RWTH Aachen)
14:00 – 14:30: Federated Learning for Network Anomaly Detection and Visual Quality Control in Shared Production – Stefanie Hittmeyer (Fraunhofer IOSB-INA)

14:30 – 15:00: Coffee break

15:00 – 17:00: Session 3: Forecasting & System Modeling

15:00 – 15:30: Inverse Modelling for Weight Prediction from Hydraulic Data – Swantje Plambeck (Hamburg University of Technology)
15:30 – 16:00: Demand Forecasting in Water Distribution Systems: A Practitioner’s Perspective on Operationalization, Transferability, and Scalability – Andreas Wunsch (Fraunhofer IOSB)
16:00 – 16:30: A Rule-Aware Prompt Framework for Structured Numeric Reasoning in Cyber-Physical Systems – Yichen Liu (Kansas State University)
16:30 – 17:00: An Algorithm for the Generation of Random System Structures – Silke Merkelbach (Fraunhofer IEM)

19:00: Dinner (self-paid, remember to bring cash, no card payment possible!) Restaurant Aposto, Monbijouplatz 12, Berlin, 10178, www.aposto-berlin.de

Day 2: 20.03.2025

08:45 – 09:00:

Welcome Coffee

09:00 – 10:30: Session 4: Ecological Impacts

09:00 – 09:30: Benchmarking Neural Architectures for Long-Horizon Forecasting in Ecological Simulations – Robin Kurth (Helmut Schmidt University)
09:30 – 10:00: AI-Driven Contextual Troubleshooting for Sustainable Fish and Plant Growth in Aquaponic Systems – Divas Karimanzira (Fraunhofer IOSB-AST)
10:00 – 10:30: Analyzing and Comparing Ecosystem Assessments of Reservoirs: A Data-Centric Approach – Christian Kühnert (Fraunhofer IOSB)

10:30 – 11:00: Coffee Break

11:00 – 12:00: Keynote: Riccardo Büttner (HSU)

12:00 – 12:45: Lunch break & Catering

12:45 – 14:15: Session 5: Robotics & Autonomous System

12:45 – 13:15: Signal Temporal Logic for Mining Guard Conditions in Hybrid System Models – Ulrike Engeln (Hamburg University of Technology)

13:15 – 13:45: Natural-Language Robot Manipulation via MCP: An Integrated Framework for Vision-Guided Pick-and-Place Automation – Daniel Gaida (TH Köln)

13:45 – 14:15: Autonomous Object Detection and Manipulation Using a Mobile Cobot – Tim Yago Nordhoff (TH Köln)

Conference Location

Fraunhofer Forum Berlin

Anna-Louisa-Karsch-Straße 2

10178 Berlin

Hosts

Important Dates

Paper Submission: ~~December 19, 2025~~ January 16th, 2026

Notification of Acceptance: February 13th, 2026

Camera-Ready Submission: March 13th, 2026

Submission Guidelines

Papers are chosen on a peer-review basis and accepted papers are published by the Helmut Schmidt University Press (openHSU) accompanied by a unique DOI. Papers with commercial character will not be taken into consideration. The length of the papers should not exceed 10 pages. Industry extended abstracts of scientific problems should be limited to 1 page.

Please use the following template for your submission:

ML4CPS template

Paper Submission will be handled via easychair:

Submission Page

For additional details and submission guidelines, please refer to

[email protected]