Deep Reinforcement Learning Based Controller Design for Model of The Vertical Take off and Landing System

In this study, the Deep Deterministic Policy Gradient (DDPG) algorithm, which consists of a combination artificial neural networks and reinforcement learning, was applied to Vertical Takeoff Landing (VTOL) system model in order control pitch angle. This algorithm selected because conventional algorithms such as Proportional Integral Derivative (PID) controllers cannot always generate suitable signal eliminating disturbance unwanted environment effects on considered system. system, training carried out for sinusoidal reference mathematical VTOL Simulink environment, through DDPG with continuous action space from deep learning methods that can produce values take structure maximize reward according determined function purpose generalization ability networks. For constant reference, tracking error performances obtained angle, is output specified were compared PID controller performance terms mean square error, integral absolute percentage overshoot settling time. The results are presented via simulations studies.