Exponential Sliding Mode Control Based on a Neural Network and Finite-Time Disturbance Observer for an Autonomous Aerial Vehicle Exposed to Environmental Disturbances and Parametric Uncertainties

The quadrotor system’s nonlinear behaviour, uncertain modelling parameters and the presence of exogenous disturbances in the surrounding environment make its flight control a critical and challenging job. Several linear and nonlinear control methodologies have been proposed so far for this purpose during the past few decades which still need improvements. In this paper, variable exponential sliding mode control (ESMC) is proposed for the attitude stabilization and altitude tracking of a quadrotor having parametric uncertainties and being exposed to exogenous disturbances. The multi-layer perceptron (MLP) neural network is incorporated with the ESMC to accommodate adaptively the effect of parametric uncertainties. The conventional weight update law of the neural network is replaced by the sliding mode effect, thereby enhancing the learning efficiency of the network without computational complexities. A finite-time disturbance observer (FTDO) is integrated with the control law to make the controller robust to exogenous disturbances and minimize the chattering issue as well. The stability of the proposed control scheme is checked and verified by the Lyapunov theory. Numerous simulations of the proposed control algorithm are carried out on a quadrotor model suffering from exogenous disturbances and parametric uncertainties in MATLAB SIMULINK and the results are compared with the SMC. The remarkable performance of the proposed control strategy demonstrates and substantiates its validity.