Abstract
This letter considers the optimization landscape of linear dynamic output feedback control with H∞ robustness constraints. We consider the feasible set of all the stabilizing full-order dynamical controllers that satisfy an additional H∞ robustness constraint. We show that this H∞ -constrained set has at most two path-connected components that are diffeomorphic under a mapping defined by a similarity transformation. Our proof technique utilizes a classical change of variables in H∞ control to establish a surjective mapping from a set with a convex projection to the H∞ -constrained set. This proof idea can also be used to establish the same topological properties of strict sublevel sets of linear quadratic Gaussian (LQG) control and optimal H∞ control. Our results bring positive news for gradient-based policy search on robust control problems.
Original language | English (US) |
---|---|
Pages (from-to) | 442-447 |
Number of pages | 6 |
Journal | IEEE Control Systems Letters |
Volume | 7 |
DOIs | |
State | Published - 2023 |
Keywords
- Hcontrol
- LQG control
- Optimization landscape
- direct policy search
- sublevel set
ASJC Scopus subject areas
- Control and Systems Engineering
- Control and Optimization