Latent Linear Quadratic Regulator for Robotic Control Tasks

TL;DR

This paper introduces LaLQR, a latent space approach that simplifies nonlinear control problems into linear-quadratic ones, enabling more efficient and generalizable robotic control.

Contribution

LaLQR is a novel method that learns a latent space where the dynamics are linear and the cost quadratic, improving efficiency over traditional MPC.

Findings

LaLQR outperforms baselines in efficiency.

LaLQR demonstrates better generalization.

LaLQR effectively imitates original MPC.

Abstract

Model predictive control (MPC) has played a more crucial role in various robotic control tasks, but its high computational requirements are concerning, especially for nonlinear dynamical models. This paper presents a $\textbf{la}$tent $\textbf{l}$inear $\textbf{q}$uadratic $\textbf{r}$egulator (LaLQR) that maps the state space into a latent space, on which the dynamical model is linear and the cost function is quadratic, allowing the efficient application of LQR. We jointly learn this alternative system by imitating the original MPC. Experiments show LaLQR's superior efficiency and generalization compared to other baselines.