Kalman-filter-based Accurate Trajectory Tracking and Fault-Tolerant Control
of Quadrotor
R. Doraiswami,Marius Brinkmann,L. Cheded
DOI: https://doi.org/10.11159/CDSR21.302
2021-05-01
Abstract:A Kalman filter(KF)-based feedforward-feedback controller is proposed using the internal model(IM)-principle for accurate tracking of a desired trajectory, and fault-tolerant control of a quadrotor, despite input and output sensor measurements being affected by unknown disturbances, measurement noise and model perturbations. The quadrotor model is unstable and nonlinear. Its input is a nonlinear function of the roll, pitch and yaw, and its output is its position in the ground-fixed coordinates. The quadrotor is subject to model uncertainties, disturbances including wind gusts, aerodynamic drags, gravitational load, and Coriolis forces, and the inputs and the outputs are affected by unknown stochastic disturbances and measurement noise. Predictive analytics is used to estimate the true input by exploiting its smoothness and the randomness of the noisy input. The nonlinear system is better approximated using the linear parameter-varying (LPV) model described by piecewise-linear Box-Jenkins model at each operating point, than by conventional approximation techniques. The system and the associated Kalman filter (KF) are identified using novel emulator-generated data by minimizing the KF residual so that identified models are accurate, consistent and reliable. The proposed tracking, fault-tolerant control, and design of the KF residuals-based design of soft sensor were successfully evaluated on a simulated laboratory-scale quadrotor.
Engineering,Computer Science