Editorial: Some recent advances in learning and adaptation for uncertain feedback control systems

Abstract

Learning and adaptation are essential abilities for feedback control systems to improve performance under uncertainties and external disturbances. In the past decades, there are more and more research interests in developing feedback controllers with learning abilities to ensure stability or optimality of closed-loop systems. In this guest editorial for the special issue, some recent advances in this area are introduced from three perspectives. The first one is about new developments in adaptive dynamic programming and reinforcement learning methods, which use function approximators such as neural networks to approximately solve the adaptive optimal control problem of uncertain nonlinear systems. The second perspective is related to the learning issues in adaptive control systems based on neural networks. The third perspective includes some new results to deal with uncertainties in feedback control systems based on traditional nonlinear control approaches such as multi-step nonlinear model predictive control and nonlinear H-1 control. Copyright © 2014 John Wiley & Sons, Ltd.