On Data-Driven Surrogate Modeling for Nonlinear Optimal Control

TL;DR

This paper explores the use of local linear time-varying models combined with iterative LQR for effective nonlinear optimal control, addressing issues in global system identification.

Contribution

It demonstrates that local LTV models with ILQR are necessary and sufficient for accurate nonlinear optimal control, overcoming global identification challenges.

Findings

Local LTV models improve control accuracy

Iterative ILQR effectively solves nonlinear control problems

Global identification suffers from high variance and ill-conditioning

Abstract

In this paper, we study the use of state-of-the-art nonlinear system identification techniques for the optimal control of nonlinear systems. We show that the nonlinear systems identification problem is equivalent to estimating the generalized moments of an underlying sampling distribution and is bound to suffer from ill-conditioning and variance when approximating a system to high order, requiring samples combinatorial-exponential in the order of the approximation, i.e., the global nature of the approximation. We show that the iterative identification of "local" linear time varying (LTV) models around the current estimate of the optimal trajectory, coupled with a suitable optimal control algorithm such as iterative LQR (ILQR), is necessary as well as sufficient, to accurately solve the underlying optimal control problem.