A Decoupled Data Based Approach to Stochastic Optimal Control Problems

TL;DR

This paper introduces a decoupled data-driven method for stochastic optimal control that combines trajectory optimization with data-learned linear controllers, demonstrating near-optimal performance on benchmark problems.

Contribution

It proposes a novel decoupled approach that integrates black-box trajectory optimization with data-driven linear control design for unknown systems.

Findings

Demonstrates near-optimal control solutions on benchmark problems.

Shows effectiveness of combining NLP-based trajectory optimization with data-driven LQR.

Validates the approach through computational examples.

Abstract

This paper studies the stochastic optimal control problem for systems with unknown dynamics. A novel decoupled data based control (D2C) approach is proposed, which solves the problem in a decoupled "open loop-closed loop" fashion that is shown to be near-optimal. First, an open-loop deterministic trajectory optimization problem is solved using a black-box simulation model of the dynamical system using a standard nonlinear programming (NLP) solver. Then a Linear Quadratic Regulator (LQR) controller is designed for the nominal trajectory-dependent linearized system which is learned using input-output experimental data. Computational examples are used to illustrate the performance of the proposed approach with three benchmark problems.