Neural network and fuzzy logic-based hybrid attitude controller designs of a fixed-wing UAV

In this paper, a mini unmanned aerial vehicle (UAV) is planned to be used in applications such as spraying pesticide and weed control in agricultural areas. According to literature review, proportional + integral + derivative (PID) structure is used to control many of these UAVs. This controller is insufficient against uncertain weather conditions and disturbance effects. In this study, many different control techniques are evaluated to select the controller structure that can respond to these uncertainties. The structure having the best result was chosen as the UAV controller. Ultrastick-25e mini UAV model is used to control the roll and yaw angle lateral dynamics. State-space presentation of the UAV longitudinal and lateral dynamics is explained, and it is just obtained for the lateral dynamics to control the attitude of the UAV under 60 km/h flight velocity condition. According to the aileron and rudder inputs, lateral dynamics simulations have successfully done by using five different controller methods such as classical PID, artificial neuro-fuzzy inference system (ANFIS), fuzzy logic controller, combined ANFIS-PID, and PD-Fuzzy-PI controllers. Moreover, three different input signals are assumed to evaluate the system response. Additionally, transient response and the time performance parameters such as overshoots, peak, rise and settling times, and steady-state error have analyzed for the designed different controllers. The simulated results for the five different controller designs showed that combined PD-fuzzy-PI and ANFIS-PID controllers have more acceptable performance than other controllers at the steady level flight condition. It is aimed that the simulation findings obtained in this study will contribute to experimental studies.