Machine learning and accelerator physics is set to converge in Liverpool from 30 March - 1 April 2026 with the announcement of RL4AA’26, the fourth Reinforcement Learning for Autonomous Accelerators workshop.
Hosted as a LIV.INNO Workshop by the University of Liverpool and the Cockcroft Institute, this event will be a forum for experts exploring the application of Reinforcement Learning (RL) to complex, real-world control systems.
RL4AA’26 is specifically designed to bring together the reinforcement learning community and the accelerator physics community, facilitating the exchange of practical insights and research results. The workshop will address the big questions facing the implementation of real-world RL, moving algorithms from simulation to practical deployment on complex, sensitive systems like particle accelerators.
The program includes tracks for talks and posters across three main areas:
- RL for Accelerators
- RL for Other Systems
- Other (covering foundational topics like sequential decision-making, RL/control theory, surrogate modelling, MLOps, adaptive control, MPC, system ID, tuning and optimisation).
The workshop promises a highly engaging and practical agenda, featuring:
- Keynote Address: Samuele Tosatto from the University of Innsbruck will deliver a keynote speech on machine learning techniques for autonomous robot learning.
- Contributed Talks: A mix of invited and contributed talks, with dedicated slots reserved for student presentations.
- Hands-on Challenge: Attendees can participate in a coding challenge focused on a real-world accelerator problem, featuring mixed-experience teams, a live Kaggle leaderboard, and awards for the top three teams.
- Facility Visit: The event will conclude on the final day with a visit to the Cockcroft Institute and the CLARA accelerator facility.
- Networking: A dedicated lunch poster session will offer ample opportunity for collaboration and discussion.
Dr. Andrea Santamaria Garcia, Lecturer in AI for Particle Accelerators at the University of Liverpool and workshop chair looks forward to welcoming participants: "We are excited to host the fourth instalment of RL4AA, which has become an essential meeting point for researchers and engineers applying advanced control to autonomous systems. The blend of high-level research and hands-on problem-solving ensures a productive and inspiring three days."
Attendance is free, but places are limited. The deadline for registration and the submission of abstracts is 31st December 2025.
For more information and to register, please visit https://indico.ph.liv.ac.uk/e/rl4aa26