A high-performance flight control approach for quadrotors using a modified active disturbance rejection technique

In practice, the parameters of the flight controller of the quadrotors are commonly tuned experimentally with respect to a certain type of reference, such as the step reference and the unit-ramp reference. In this way, the performance of the flight controller might be affected by the variations of the references in real-time flights. Besides, real-time dynamic effects such as measure noises, external disturbances and input delays, which are usually neglected in the reported works, could easily deteriorate the performances of the flight controllers. This work is thereby motivated to develop a high-performance flight control approach utilizing a modified disturbance rejection technique for the quadrotors suffering from input delays and external disturbances. This control approach is developed in a cascaded structure and the attitude angles are chosen as the pseudo control inputs of the translational flight of the quadrotors. To facilitate the development, the dynamic model of the quadrotors is firstly formulated by including the effects of input delays, and the dynamics of the pseudo control variables are identified through real-time experiments. Based on the identified model, the flight control approach is proposed with a modified active disturbance rejection technique, which consists of a time optimal tracking differentiator, an extended state observer/predictor, and a nonlinear proportional-derivative controller. The tracking differentiator is designed to generate smooth transient profiles for the references, and the extended state observer/predictor is implemented for lumped disturbance estimation and state estimation considering the input delays. With the aid of the tracking differentiator and the extended state observer/predictor, the nonlinear proportional-derivative controller can thereby establish a fast tracking control and effectively reject the estimated disturbances. To verify the feasibilities of this development, comparative tests are carried out in both simulations and experiments. The results show that in the presence of small lumped disturbances, such as the measurement zero-drift, the steady-state errors of the proposed control approach for the ramp responses are less than 2 cm, and in the tests of sinusoidal trajectory tracking, the cross-tracking errors are less than 0.04 m. When with large disturbance airflow that is equivalent to strong breeze, the steady-state error achieved by the proposed flight controller is also less than 10 cm. All of these facts demonstrate the effectiveness of this development. (C) 2016 Elsevier B.V. All rights reserved.