Efficient model-based reinforcement learning for approximate online optimal

TL;DR

This paper introduces an efficient online reinforcement learning approach for nonlinear control systems using the StaF kernel method, which approximates value functions locally, reducing computational complexity while maintaining stability and near-optimality.

Contribution

The paper presents a novel local approximation method for online optimal control, reducing basis function requirements compared to traditional global methods.

Findings

Achieves stability and near-optimal control with fewer basis functions.

Demonstrates effectiveness through simulation results.

Provides a scalable approach for nonlinear control systems.

Abstract

In this paper the infinite horizon optimal regulation problem is solved online for a deterministic control-affine nonlinear dynamical system using the state following (StaF) kernel method to approximate the value function. Unlike traditional methods that aim to approximate a function over a large compact set, the StaF kernel method aims to approximate a function in a small neighborhood of a state that travels within a compact set. Simulation results demonstrate that stability and approximate optimality of the control system can be achieved with significantly fewer basis functions than may be required for global approximation methods.