Geometric approach enhances robust least-squares optimization for data-driven predictive control

This paper introduces a novel geometric approach to robust least-squares optimization, extending classical methods by modeling data uncertainty as a metric ball on a Grassmannian manifold. This formulation leads to a min-max problem that can be solved efficiently, offering a clear geometric interpretation. When applied to data-enabled predictive control for linear-quadratic tracking, the proposed method demonstrates improved robustness and better scalability compared to existing robust least-squares techniques, particularly under conditions of small uncertainty. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a novel optimization technique that could enhance the performance and robustness of AI-driven control systems.

RANK_REASON This is a research paper published on arXiv detailing a new optimization method.

Read on arXiv cs.LG →

• Shreyas Bharadwaj
• arXiv
• Grassmannian manifold

• paper
• other