学术报告第202162期

报告题目：Reinforcement Learning-Based Control of Uncertain Nonlinear Systems

报 告 人：谢立华，新加坡南洋理工大学教授

报告时间：2021年11月30日（周二）上午10:30-11:30

报告地点：南一楼中314

报告摘要:

Reinforcement learning (RL), inspired by learning behaviour in nature, is a goal-oriented learning strategy wherein the agent learns the policy to optimize a pre-defined reward by interacting with the environment. For being data-driven, effectiveness in reaching optimal behavior, and adaptiveness to uncertain environment, RL has undergone rapid development in control community. In this talk, we shall first discuss RL based disturbance rejection control for uncertain nonlinear systems with known nominal part. An extended state observer is first designed to estimate the system state and the total uncertainty. Based on the output of the observer, the control compensates for the total uncertainty in real time, and simultaneously, online approximates the optimal policy for the compensated system using a simulation of experience based RL technique. The approach does not require PE condition or probing signals. We then extend the study to systems with unknown nominal part, where a novel concurrent adaptive extended observer is developed to jointly estimate the parameters of the systems and the state, and a simulation of experience based RL is used to approximate the optimal policy.

报告人简介：

谢立华，新加坡南洋理工大学电气与电子工程学院教授，新加坡工程院院士，IEEE、IFAC和CAA会士。曾任南洋理工大学控制与仪表系主任。研究领域包括鲁棒控制、网络控制、定位与无人系统。曾出版了9本书，480多篇期刊文章，20项专利和技术披露，Thomson Reuters和Clarivate Analytics 高被引作者 (2014-2020)。他目前是《无人系统》杂志的主编和中国科学-信息科学杂志的副主编，曾是IET控制系列丛书主编，IEEE Transactions on Automatic Control, Automatica, IEEE Transactions on Control System Technology, IEEE Transactions on Control of Network Systems 等杂志的副主编。同时也是IEEE杰出讲师（2011-2014）, IEEE控制系统学会理事（2016-2018）and CDC2023总主席。