Inverse Optimal Control for High‐Order Nonlinear Fully Actuated Systems With Unknown Reward Weights

ABSTRACT Optimal control of high‐order nonlinear fully actuated systems (FASs) has garnered considerable attention in control theory. However, most existing studies assume that the weights in the performance index are either known or pre‐specified. To overcome the challenge posed by unknown weights, we propose an inverse optimal control (IOC) approach to learn the reward weights by observing the trajectory data of an expert system. Compared with conventional optimal control methods, our approach not only enables faster system stabilization but also achieves superior performance. Theoretically, we rigorously prove the convergence of the algorithm and the stability of the closed‐loop system, thereby establishing a theoretical underpinning for the method's reliability. Simulation results demonstrate that the proposed IOC method accurately learns the weight parameters of the expert system and achieves control performance comparable to that of the expert policy across different tasks.