Simon Hirländer
Simon Hirländer is Vice Director of the IDA Lab and leads the Smart Analytics and Reinforcement Learning (SARL) team at Paris Lodron University Salzburg, where he also holds a postdoctoral position. He serves as a Visiting Scientist at CERN and is known for applying reinforcement learning and model predictive control to particle accelerator optimization, including pioneering work on safe, uncertainty-aware RL agents for autonomous accelerator control. His research spans model-based and hierarchical reinforcement learning, Gaussian process MPC, and causal inference, with applications extending to industrial control problems.
“Rather than evaluating safety purely online, which is often computationally prohibitive for the high-frequency control loops required in modern CPSs, our approach uses offline MPC as an oracle to pre-define a 'feasible state-action space' (F)”
Source→“Pre-computing the safe boundary offline solves this by ensuring the agent never crosses the point of no return in the first place”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.