Sum of Squares Polynomial Approximate Dynamic Programming for the Ergodic Problem

• 成果类型: Article
• 署名作者: Hexner, Gyorgy
• 署名单位: RAFAEL ADVANCED DEFENSE SYSTEMS
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2025.3559462
• 发表日期: 2025
• 页码: 6253-6259
• 关键词: Polynomials; iterative methods; Approximation algorithms; Closed loop systems; dynamic programming; Output feedback; optimal control; mathematical models; Stability criteria; minimization; Approximate Dynamic Programming; ergodic systems; stochastic control

摘要

The purpose of this note is to extend the approximate dynamic programming (ADP) method to the infinite time stochastic optimal control (ergodic) problem. It is also shown that a modification of the basic algorithm solves the output feedback problem. The ADP solution of the ergodic problem differs in significant ways from its deterministic counterpart.