The Fragility of Learning LQG Controllers

TL;DR

This paper establishes fundamental lower bounds on the sample complexity for learning LQG controllers from data in partially observed systems, highlighting the challenges posed by system fragility.

Contribution

It derives information-theoretic lower bounds for offline learning of LQG controllers, linking system fragility to high sample complexity in partially observed control.

Findings

Fragile robust control problems require many samples to learn effectively.

Certainty-equivalent methods are asymptotically optimal under certain conditions.

System fragility impacts the difficulty of data-driven control in classical examples.

Abstract

Learning methods are increasingly used to synthesize controllers from data, yet existing sample-complexity characterizations for continuous control are sharp only in the fully observed setting. This paper studies the partially observed case by deriving information-theoretic lower bounds for learning Linear Quadratic Gaussian (LQG) controllers from offline trajectories generated by a (linear) exploration policy. We prove an $\varepsilon$-local minimax excess-cost lower bound that applies to any algorithm mapping the offline dataset to a stabilizing linear controller. The bound is expressed in terms of the Hessian of the LQG cost with respect to model parameters and the inverse Fisher Information induced by the exploration policy. We further provide system-theoretic characterizations of these objects, enabling transparent construction of hard instances. Instantiating the bound on classical fragile robust-control examples, including variants of the Doyle LQG fragility counterexample and non-minimum-phase systems, demonstrates when fragile robust control problems translate into high sample complexity for learning-enabled control. These results suggest the asymptotic optimality of certainty-equivalent synthesis and motivate the importance of both task-directed experiment design and system co-design for sample-efficient learning in partially observed control.