Robust attitude control of a quadrotor unmanned aerial vehicles using optimized disturbance observer

Abstract

This article proposes a method of realizing robust attitude stabilization control of a quadrotor UAV using the disturbance observer (DOB), where Q-filter is optimally designed considering performances of attenuating disturbance and estimation noise, and robust stability against parameter variations and sampling duration delay due to discrete time control in microcontroller. Disturbance rejection performance mainly depends on the bandwidth determined by cut-off frequency of DOB's Q-filter. However, extension of the Q-filter's bandwidth must be restricted due to the requirement of robust stability against parameter uncertainty and sampling duration delay in onboard microcontroller for flight control of quadrotor. In order to optimize the disturbance rejection performance under the restriction of bandwidth, an H ∞ norm minimization problem is defined considering the frequency bands of disturbance and noise. The optimization problem is transformed into standard H ∞ control design problem by pseudo loop factorization method so as to resolve the structural constraint of Q-filter. Then, stability of the discrete time system configured by digital controller is tested in relation with the time constant of the designed DOB's Q-filter under different inertia moment values. The comparisons between simulation results by different control methods validate that the robust control method using proposed DOB provides a better performance of disturbance attenuation than that using conventional DOB.

Parameters self‐turning controller fused with an adaptive nominal model for disturbance observer: Application to direct drive manipulator with significant payload changes

“…14,15 The disturbance observer (DOB) has proven to be a highly effective tool for robust control, aiding in the estimation of external disturbances and uncertainties in the dynamics of the plant. [16][17][18][19][20][21][22][23] These disturbances are estimated by comparing the output deviations of the plant with a nominal model, and they are subsequently compensated at the input to mitigate their effects. At present, DOB has been widely used in the controller design of the traditional manipulator, and the results are satisfactory.…”