Font Size: 
Performance Evaluation of Parametric Estimators with Application to Quadrotors
Mikhail Kakanov, Stanislav Tomashevich, Vladislav Gromov, Oleg Borisov, Fatimat Karashaeva, Anton Pyrkin

Last modified: 2023-06-29


This paper addresses the problem of quadrotor parameters estimation. The dynamic model of the quadrotor includes parameters, which may be either unknown, or just partially known or even time-varying. In this regard, the parametric identification problem may arise during the control design. There are a range of different online and offline estimation approaches, which provide a solution to that problem under various assumptions. However, they all have certain pros and cons and different transient response characteristics. The goal of the study is to evaluate the performance of several estimation approaches applied to the quadrotor model. Estimation design includes the parametrization of the 6-DOF quadrotor model through the linear regression and implementation of estimators based on the following approaches: ordinary least squares, advanced Kalman filter, and dynamic regressor extension and mixing (DREM). Estimation performance is conducted employing simulations and experiments. Discussions on the obtained results with the detailed comparison analysis of the considered approaches are provided.