LQGNet: Hybrid Model-Based and Data-Driven Linear Quadratic Stochastic Control

TL;DR

LQGNet is a hybrid stochastic control model that combines classical LQG principles with data-driven learning to effectively handle partially known system dynamics, outperforming traditional methods.

Contribution

It introduces LQGNet, a novel hybrid control framework that integrates model-based and data-driven approaches for stochastic control with partially known dynamics.

Findings

LQGNet outperforms classic stochastic control in mismatched system scenarios.

The model effectively learns to cope with partially known system dynamics.

LQGNet maintains the structure of traditional LQG control while enhancing robustness.

Abstract

Stochastic control deals with finding an optimal control signal for a dynamical system in a setting with uncertainty, playing a key role in numerous applications. The linear quadratic Gaussian (LQG) is a widely-used setting, where the system dynamics is represented as a linear Gaussian statespace (SS) model, and the objective function is quadratic. For this setting, the optimal controller is obtained in closed form by the separation principle. However, in practice, the underlying system dynamics often cannot be faithfully captured by a fully known linear Gaussian SS model, limiting its performance. Here, we present LQGNet, a stochastic controller that leverages data to operate under partially known dynamics. LQGNet augments the state tracking module of separation-based control with a dedicated trainable algorithm. The resulting system preserves the operation of classic LQG control while learning to cope with partially known SS models without having to fully identify the dynamics. We empirically show that LQGNet outperforms classic stochastic control by overcoming mismatched SS models.