Neural network-driven adaptive PD controller for hexacopter UAV path tracking

Abstract

Unmanned Aerial Vehicles (UAVs), particularly hexacopters, have demonstrated remarkable potential for a variety of applications, including military operations, surveillance, precision agriculture and disaster response. However, their deployment is often hindered by challenges arising from their nonlinear dynamics, external disturbances, system uncertainties, and the limitations of conventional control approaches. To address these challenges, this study proposes an adaptive proportional-derivative (PD) controller augmented with a neural network (NNAPD) for hexacopter trajectory tracking. The proposed system incorporates an online tuning mechanism for PD parameters using the neural network, enabling real-time adaptability to environmental and system changes. This approach ensures enhanced robustness, precise maneuverability, and consistent performance under dynamic and unpredictable operating conditions. The effectiveness of the proposed method is validated through simulations, demonstrating its ability to maintain stability and achieve accurate trajectory tracking in the presence of system uncertainties and disturbances.