PAC-Bayesian Optimal Control with Stability and Generalization Guarantees

TL;DR

This paper introduces a PAC-Bayesian framework for stochastic nonlinear optimal control that guarantees stability and generalization, effectively balancing empirical performance with prior knowledge, especially under limited data conditions.

Contribution

It develops a novel PAC-Bayesian approach for control design that provides generalization bounds and stability guarantees, incorporating prior knowledge and neural controllers.

Findings

Provides rigorous generalization bounds for SNOC.

Demonstrates stability guarantees with neural controllers.

Shows improved control reliability in simulated robotics tasks.

Abstract

Stochastic Nonlinear Optimal Control (SNOC) seeks to minimize a cost function that accounts for random disturbances acting on a nonlinear dynamical system. Since the expectation over all disturbances is generally intractable, a common surrogate is the empirical cost, obtained by averaging over a finite dataset of sampled noise realizations. This substitution, however, introduces the challenge of guaranteeing performance under unseen disturbances. The issue is particularly severe when the dataset is limited, as the trained controllers may overfit, leading to substantial gaps between their empirical cost and the deployment cost. In this work, we develop a PAC-Bayesian framework that establishes rigorous generalization bounds for SNOC. Building on these bounds, we propose a principled controller design method that balances empirical performance and prior knowledge. To ensure tractability, we derive computationally efficient relaxations of the bounds and employ approximate inference methods. Our framework further leverages expressive neural controller parameterizations, guaranteeing closed-loop stability. Through simulated examples, we highlight how prior knowledge can be incorporated into control design and how more reliable controllers can be synthesized for cooperative robotics.