News & Highlights

Invited Talk at NeurIPS 2026 Workshop (Paris)

Invited talk on "Koopman Stabilised Symplectic World Models for Safe Offline-to-Online RL in Particle Accelerators" at the NeurIPS 2026 Workshop RL for Experimental Sciences: Bridging the Simulation-to-Reality Gap in Paris. Presenting data-driven control, spectral Koopman bounds, and physics-informed offline RL for large-scale physical infrastructures.

Invited Talk at NexusNow powered by COPA-DATA

Invited talk "Daten, die handeln dürfen: Wie Agentic AI und zenon die industrielle Prozessführung verändern" at NexusNow in Salzburg, on how agentic AI and reinforcement learning can take over process control in industrial automation.

Two DEXA 2026 Workshop Papers (AI4IP)

Two papers with FH Salzburg in the DEXA 2026 Workshops (Springer CCIS): "Anticipatory Reinforcement Learning for Trajectory Tracking" and "Reinforcement Learning for Optimal Experiment Design in Parameter Identification of Mechatronic Systems".

EPJ Research Infrastructures: Automating Accelerator Tuning at GSI/FAIR

Journal article "Automating Accelerator Tuning at GSI/FAIR" (EPJ Research Infrastructures 10, art. 23). Automated tuning cut SIS18 injection losses from 45% to 12%; my team contributed the data-driven Gaussian Process MPC.

RL4AA'26 Workshop, Liverpool

Member of the Scientific Organising Committee of the 4th RL4AA collaboration workshop at the University of Liverpool / Cockcroft Institute. Gave the tutorial "Introduction to Deep RL Agents", ran the challenge session and joined the discussion panel; the SARL team presented six posters (Causal GP-MPC, Koopman world models, RL beyond greedy optimisation, SIS18 GP-MPC, and more).

Two EUROCAST 2026 Papers

Two contributions with FH Salzburg accepted and presented at EUROCAST 2026: "Safe Reinforcement Learning using Ideas from Model Predictive Control" (arXiv:2607.07252) and "Sample-Efficient Pareto Front Modeling for Energy-Aware Reinforcement Learning using Bayesian Optimization" (arXiv:2607.03140).

Panel Discussion: B&R / ABB WeGrow Day

Panelist at B&R Industrial Automation (ABB Group) company-wide "WeGrow Day" in Eggelsberg, alongside CEO Florian Schneeberger. Topics included ROI of AI in industrial automation, physics-informed AI as competitive advantage, and human-in-the-loop risk management.

DIH West Workshop: Agentic AI for Industry

Led the workshop "Bessere Entscheidungen durch agentische KI" at Innovation Salzburg as part of the DIH West programme, targeting SMEs in Western Austria.

S2 #17 | Wie Reinforcement Learning Unternehmen verändert

Künstliche Intelligenz ist längst im Unternehmensalltag angekommen – doch was kommt nach Chatbots? In dieser Folge von EASY cheesy DIGITAL sprechen wir über einen der spannendsten KI-Ansätze der Zukunft: Reinforcement Learning.

Im Gespräch mit Moderator Rudolf Grimm erklärt Simon Hirländer, wie Algorithmen lernen unter Unsicherheit gute Entscheidungen zu treffen und welches enorme Potenzial darin für Unternehmen liegt.

Dieses Projekt ist eine Initiative des DIH West, finanziert durch das Bundesministerium für Arbeit und Wirtschaft und das Bundesministerium für Klimaschutz, Umwelt, Energie, Mobilität, Innovation und Technologie im Rahmen des FFG Programms „Digital Innovation Hubs in Österreich“. Foto: Benedikt Schemmer.

3rd International RL Bootcamp

Co-organizing the 3rd Reinforcement Learning Bootcamp at Unipark Nonntal, University of Salzburg. New: custom Gym environments for beginner fundamentals (DP, Q-learning, SARSA).

IPAC 2026: 4 Accepted Contributions

Presenting four works at IPAC 2026 in Deauville, France:

• RL Beyond Greedy Optimisation (Poster, Peer-Reviewed Track) - RL for delayed-consequence regimes in AWAKE.
• Robust Real-Time Optimization of SIS18 Injection (Poster, Light Peer Review) - GP-MPC for robust injection.
• Causal GP-MPC (Poster) - Embedding structure, safety, and online learning.
• Koopman-Stabilised World Models (Poster) - Offline RL using Koopman theory.

HB2025: Contributed Oral (China)

Contributed oral paper "Geoff Framework for Automated Accelerator Tuning" (FRCAA01) at the 71st ICFA Advanced Beam Dynamics Workshop in Huizhou, China. Key result: SIS18 injection losses reduced from 45% to 15% using first multi-objective/multi-fidelity Bayesian optimization. Published by JACoW.

Appointed Vice Director of IDA Lab Salzburg

Appointed Vice Director of the Intelligent Data Analytics (IDA) Lab at the University of Salzburg, succeeding Roland Kwitt, alongside Director Wolfgang Trutschnig.

New Doctoral Programmes: KI-MINTFIT and DSP Human-Centered AI

Start of the BMBWF-funded doctoral programme KI-MINTFIT with PH Salzburg (Co-Investigator) and approval of the doctoral college "Human-Centered AI: From Data & Algorithms to Interfaces" (2026-2029), where I am a founding member of the supervision team. The faculty also authorised me as main supervisor of doctoral theses.

New Industry Projects

New contract research led by the SARL team: FOCUS with COPA-DATA (industry-funded PhD, energy forecasting and optimisation), RL-PD with W&H Dentalwerk (RL for piezo drives), DS STIWA II with STIWA, SCOPE with KACO, KATIE with eurofunk Kappacher (AI analysis of emergency calls), and a cooperation with DLR on RL-based rocket engine control.

Quantum Science and Technology: RL for Quantum Gates

Journal article "Achieving fast and robust perfect entangling gates via reinforcement learning" (Quantum Science and Technology 11, 015030), with L. Grech, C. P. Koch and colleagues.

TEDxSalzburg Salon: "Rethinking Economy"

Invited speaker at the TEDxSalzburg Salon hosted at B&R Industrial Automation (ABB Group) in Eggelsberg, discussing the intersection of AI and economic transformation.

Keynote at zenonIZE 2025

Delivered keynote with Lucas Punzenberger (COPA-DATA) on "Future of AI-driven automation and the path to more intelligent self-learning systems in industry" at zenonIZE 2025 in Salzburg Mozarteum Foundation.

INSPIRE Project - AI in Education

Partner in the INSPIRE project (Intelligent Novel Support for Personalized Instruction and Robust Evaluation in STEM Lessons at Primary School), funded by the State of Salzburg's "AI Research Salzburg 2025" initiative, led by Christina Egger (PH Salzburg).

2nd RL Bootcamp

Co-organizing the 2nd Reinforcement Learning Bootcamp at the University of Salzburg. The bootcamp features hands-on workshops on Advanced Policy Gradients, Actor-Critics, and Continuous Control.

Keynote at iDSC'25

Delivered keynote "From Theory to Practice: Reinforcement Learning as the 21st-Century ML Paradigm" at the 6th Interdisciplinary Data Science Conference.

RL4AA Workshop Hamburg: Session Chair

Co-organized the RL for Autonomous Accelerators workshop in Hamburg. Served as Session Chair for Tuesday Session 1 and hosted the closing discussion (polls, word clouds, panel format).

RL4AA 2025 Contributions

Contributions to the Workshop on RL for Autonomous Accelerators:

• [Talk] Reliable Control via Contextual MPC (S. Hirlaender, S. Trausner)
• [Poster] Robust Real-Time Optimization of SIS18 (S. Appel et al.)
• [Poster] Causal GP-MPC (O. Mironova et al.)
• [Poster] Curriculum-Guided PPO (F. Matanza et al.)

🏆 IEEE INDIN'24: Best Presentation Award

Received the Best Presentation Award at the IEEE International Conference on Industrial Informatics (INDIN) 2024 for the paper "Comparison of MPC and PPO for a 1-DOF Helicopter System" (G. Schäfer, J. Rehrl, S. Huber, S. Hirlaender).

1st RL Bootcamp

Successfully organized the first Reinforcement Learning Bootcamp at the University of Salzburg with participants from academia and industry.

TEDx Talk: "AI is what we make it"

Delivered TEDx talk on "AI is what we make it" at TEDxSalzburg, discussing the future of artificial intelligence and its societal impact.

Invited Session: SMPS 2024 (Springer)

Invited to organise and present at a special session on "Python-Based Reinforcement Learning on Simulink Models" at the SMPS 2024 (International Conference on Soft Methods in Probability and Statistics), published by Springer.

Invited Colloquium: ScaDS.AI Leipzig

Delivered invited colloquium talk at the ScaDS.AI (Center for Scalable Data Analytics and Artificial Intelligence) at Universität Leipzig on reinforcement learning for autonomous systems.

RL4AA Workshop Salzburg

Hosted the RL for Autonomous Accelerators workshop in Salzburg, establishing a key platform for international collaboration.