Learning-Based Control of Continuous-Time Systems Using Output Feedback

This paper presents an adaptive optimal control approach for continuous-time linear systems with output feedback. The method fills in the gap in the literature of reinforcement learning and adaptive dynamic programming that has been focused exclusively on either discrete-time systems or continuous-time systems with full-state information. The approach utilizes the historical continuous-time input-output trajectory to reconstruct the current state, without discretizing the system dynamics or using a state observer. By exploiting the policy iteration method, suboptimal output-feedback controllers can be directly obtained from collected input-output trajectory data in the absence of an accurate dynamic model. The effectiveness of the proposed learning-based PI algorithm is demonstrated through a practical example of F-16 aircraft control.