Neural Identification for Control

TL;DR

This paper introduces a neural network-based control method that learns to stabilize unknown nonlinear systems by jointly identifying system dynamics and control laws using a Lyapunov stability framework.

Contribution

It presents a novel self-supervised learning approach that integrates system identification and control law synthesis for nonlinear systems using neural networks.

Findings

Successfully stabilizes various nonlinear systems.

Demonstrates effective trajectory tracking and balancing.

Utilizes Lyapunov theory for stability guarantees.

Abstract

We present a new method for learning control law that stabilizes an unknown nonlinear dynamical system at an equilibrium point. We formulate a system identification task in a self-supervised learning setting that jointly learns a controller and corresponding stable closed-loop dynamics hypothesis. The input-output behavior of the unknown dynamical system under random control inputs is used as the supervising signal to train the neural network-based system model and the controller. The proposed method relies on the Lyapunov stability theory to generate a stable closed-loop dynamics hypothesis and corresponding control law. We demonstrate our method on various nonlinear control problems such as n-link pendulum balancing and trajectory tracking, pendulum on cart balancing, and wheeled vehicle path following.