Data-Driven Discovery of Unmanned Aerial Vehicles Dynamics*

This paper presents Wide-Array of Nonlinear Dynamics Approximation (WyNDA) as a method for discovering the governing equations of Unmanned Aerial Vehicles (UAVs). WyNDA employs an array of basis functions to effectively capture the intricate dynamics inherent in UAVs. In order to estimate the structure and parameters of the governing equation, WyNDA utilizes a discrete-time exponential forgetting factor observer. To validate the effectiveness of this method, we conducted numerical simulations, with a particular focus on the path-following behavior of a quadcopter UAV. Our simulations demonstrate that WyNDA displays high accuracy and robustness in deciphering the system dynamics of quadcopter UAVs, even when dealing with limited datasets. These findings emphasize the potential for real-time implementation and underscore the broader applicability of our approach across diverse dynamical systems.