A Perspective Analysis on Effects of Varying Inputs on UAV Model Estimation

The aviation use of Unmanned Aerial Vehicles (UAVs) necessitates a strong control design. The key to designing a durable control system is a well-developed flight dynamics model. System-based model identification techniques provide a useful means for estimating UAV modal parameters thereby saving time and money. Although considerable research is performed on UAV flight dynamics analysis utilizing system identification techniques, limited work exists that compares various techniques of system identification for UAVs in an exhaustive manner. Moreover, the research contributions toward performance evaluation and comparison of system identification methods outputs are even more scarce, especially under varying environmental conditions. In this study, a comprehensive framework utilizing various linear and nonlinear estimation techniques estimates unknown UAV model dynamics. To analyze the effects of varying flight conditions, a detailed analysis is performed which includes a parametric sweep of environmental elements as input arguments to the estimation process. The comparison involves the estimation of key performance parameters such as residual analysis, final prediction error, and fit percentages. Through rigorous analysis, it is demonstrated that the proposed framework predicts system parameters under a variety of conditions, thereby confirming its validity. It has also been demonstrated that a parametric sweep of environmental conditions, can be utilized to improve the authenticity of models’ data learning ranges, and their response to the prediction parameters. This paper, to the best of our knowledge, provides an elaborate platform for researchers to carry out comprehensive model prediction under a wide range of environmental conditions.