Research
My research focuses on Reinforcement Learning (RL) and its application to complex, real-world systems. I lead the Smart Analytics & Reinforcement Learning (SARL) team, bridging the gap between theoretical advancements and industrial applications.
Core Research Areas
Learning from limited data for physical systems
Constrained optimization for critical infrastructure
Optimization of manufacturing & energy systems
AI for particle accelerators (CERN, DESY, etc.)
Current Projects
Forecasting and optimization under constraints and uncertainty for sustainable industrial energy systems
Principal Investigator. Contract research project in collaboration with Ing. Punzenberger COPA-DATA GmbH. Developing advanced RL and optimization methods for managing industrial energy systems under uncertainty.
Principal Investigator Energy Industrial AIIntelligent Novel Support for Personalized Instruction and Robust Evaluation in STEM Lessons
Principal Investigator (PLUS partner). Application-oriented basic research project funded by Land Salzburg (WISS2030) in collaboration with PH Salzburg. Focusing on AI-driven personalized learning and evaluation in primary education.
Principal Investigator Education AI for GoodSmart Component Process Evaluation
Co-Lead. Research project focusing on smart component evaluation and process optimization.
Industrial AI Process EvaluationThe Gamma Factory is an ambitious CERN initiative to create a "super light source" by colliding laser pulses with partially stripped ions in the LHC. My involvement focuses on the accelerator physics challenges of maintaining stable ion beams and optimizing the collision luminosity, leveraging the same advanced control techniques used for the standard LHC runs.
Researching Multi-Agent Reinforcement Learning (MARL) applications in 5G/6G wireless communication. Focusing on optimizing UAV-assisted networks to minimize information age (AoI) and energy consumption, enabling autonomous, self-optimizing aerial network swarms.
Regelungsoptimierung von Piezoantrieben
Optimization of piezoelectric device drives.
Medical Control Theory PiezoData Science STIWA II
Research project with STIWA Group. Focusing on assembly line optimization using Hierarchical Reinforcement Learning to manage complex, multi-stage decisions in high-speed manufacturing.
Manufacturing Hierarchical RL OptimizationNautic Robotics Occupation Maps
Research cooperation with EvoLogics GmbH and Paris Lodron University Salzburg (PLUS) / Dept. of AI & Human Interfaces. Transferring control theory expertise to autonomous maritime systems, leveraging similarities between fluid dynamics and beam physics.
Funding: EvoLogics GmbH (contract research) · Duration: Oct 2024 – Mar 2025 · Team: S. Hirländer (PI), C. Borgelt, S. Dabarera
Robotics Maritime Control Theory IndustryDoctoral Research
My doctoral research focused on the derivation of exact solutions for indirect transverse field effects in elongated structures, with direct applications to the CERN Large Hadron Collider (LHC) and Proton Synchrotron (PS). This work addressed the critical challenge of Indirect Space Charge Driven (ISCD) effects, which serve as a fundamental limit to beam stability and intensity.
I developed a novel theoretical framework utilizing complex Green functions to accurately model ISCD tune-shifts, providing the first comprehensive explanation for intensity-dependent phenomena observed in the PS Multi-Turn Extraction. Key contributions include the derivation of closed-form solutions for magnetic interactions in accelerator components and the development of new operators for tune-shift estimation. These models provide essential insights for mitigating beam instabilities in current operations and are scalable for future projects like the High-Luminosity LHC (HL-LHC).
Accelerator Physics CERN Green Functions LHCCompleted Projects
DeepREL
Deep Reinforcement Learning for Large and Complex Systems. International collaboration with University of Malta and CERN. Solving complex control problems for the LHC.
CERN LHCDanieli X – PLUS
Research collaboration with Danieli Automation focusing on applied research in steel automation.
Steel IndustryDEOP 2.5
Dynamische Energieoptimierung. Applied research project with COPA-DATA GmbH focusing on dynamic energy optimization.
EnergyUNSAD
Unsupervised Anomaly Detection in Sterilisation Devices. Research collaboration with W&H Dentalwerk Bürmoos GmbH.
Medical Anomaly DetectionAREP
Automatische Rechnungsprüfung. Applied research project with JUIX GmbH focusing on automated invoice auditing.
FinTechMuMa II & III
Mustererkennung Maschinendaten. Research collaborations with Palfinger AG on pattern recognition in machine data.
ManufacturingRedlink
Research co-operation with Redlink GmbH exploring the intersection of Neuro-Symbolic AI and Knowledge Graphs.
Neuro-Symbolic AI Knowledge GraphsDS STIWA
Data Science STIWA. Project Researcher (Lead: Trutschnig).
ManufacturingDEOP 2.0
Dynamische Energieoptimierung 2.0. Project Lead.
EnergyLLMs for Enterprise
Project Lead. Exploring Large Language Models for enterprise applications.
LLM Enterprise AICollaborations
I actively collaborate with leading research institutions and industry partners:
Stylolites are rough surfaces formed by pressure solution in rocks. This research applied statistical physics and non-linear dynamics to model their formation and morphology. By treating rock deformation as a complex system, we derived scaling laws that describe the roughness of these geological features, drawing parallels to crack propagation and growth processes in physics.