Integral Reinforcement Learning for Optimal Tracking

InIntegral Reinforcement Learning (IRL) controlReinforcement Learning (RL) system design, a common objective is to find a stabilizing controller that ensures the system’s output tracks a desired reference trajectoryReference trajectory. Optimal controlOptimal control theory aims to achieve this goal by determining a control law that not only stabilizes the error dynamics but also minimizes a predefined performance indexPerformance index. Reinforcement learningReinforcement Learning (RL) (RL) algorithms have proven to be effective in solving the optimal tracking control problemOptimal Tracking Control Problem (OTCP) (OTCP) for both discrete-time (Dierks and Jagannathan 2009; Wang et al. 2012; Modares et al. 2014) and continuous-time systems (Zhang et al. 2011). RL algorithms not only learn optimal tracking control solutions but also stabilize the tracking errorTracking error systems.