Reproduce “Generalizable Output Feedback Optimal Control via Transformers Without Explicit State Estimation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Maitham F. AL-Sunni, Turki Bin Mohaya, John M. Dolan, Peter Seiler
The transformer imitates optimal LQG control across simulated partially observed linear systems using recent measurements and cost and noise descriptions rather than plant matrices or an explicit state estimate. It is trained on optimal controller rollouts. Across 28 systems, median relative sub-optimality was 0.4% on training systems and 1.0% on unseen systems after fine-tuning; these are rollout results, not stability guarantees.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.