Parameter Identification Techniques for UAV Aerodynamic Models: A Systematic Literature Review

Abstract

The utility of any mathematical model for UAV flight is fundamentally constrained by the accuracy of its underlying physical parameters, such as moments of inertia, aerodynamic derivatives, and motor constants. This systematic literature review (SLR) offers an exhaustive evaluation of methodologies used to identify these parameters, synthesizing research from over 180 peer-reviewed publications. We categorize the state-of-the-art into three primary methodological streams: static experimental characterization, offline batch estimation, and online recursive identification. Static methods, including wind tunnels and pendulum-based inertia measurement, are analyzed for their precision versus high cost. Offline batch methods, such as Least Squares and Maximum Likelihood Estimation, are reviewed for their ability to process large datasets to extract global parameter sets. The review places special emphasis on online recursive identification, where techniques like Extended Kalman Filters (EKF) and Model Reference Adaptive Systems (MRAS) update parameter estimates in real-time. This is critical for UAVs experiencing significant state changes, such as those carrying liquid payloads or undergoing structural damage. Each technique is assessed based on sensitivity to sensor noise, requirement for persistent excitation, and computational complexity. Our analysis reveals a shift toward “Grey-Box” identification, where machine learning learns the residuals of physical models to capture complex aerodynamic effects. By providing a comprehensive taxonomy and comparative performance analysis, this SLR serves as a definitive guide for researchers tasked with the calibration of high-fidelity UAV simulators and adaptive control systems.