Design and implementation of a cascaded MPC with a dynamic battery-aware auxiliary reference for unmanned aerial vehicles

Abstract

Unmanned Aerial Vehicles (UAVs) have been a focus of interest within researchers and robotics industry due to their versatility and potential to improve a wide range of applications. Nevertheless, modeling control of UAVs, especially quadrotors, remains difficult due to the significant link between position and orientation dynamics. Although conventional control strategies are widely used for their simplicity and ease of implementation, they have a poor performance in practical applications, especially under uncertain operating conditions. For this reason, numerous advanced control strategies have been implemented to improve trajectory tracking. However, drone efficiency in practical tasks is also affected by flight autonomy, and these algorithms do not consider energy consumption rates. Moreover, even though battery management systems (BMS) have been designed to prevent overdischarges, they do not consider voltage sags caused by thrust variations. In this thesis, these issues are addressed through the implementation of a proposed control structure that consists of two main components: a model predictive control (MPC) strategy and a dynamic battery-aware auxiliary reference (DBAR) system. First, the design and implementation of a hierarchical cascaded MPC for three-dimensional trajectory tracking is presented. In contrast to previous research, the heavy computational burden is managed by decomposing the overall MPC strategy into two different schemes. The first scheme controls the translational displacements, while the second scheme regulates the rotational movements of the quadrotor. On the other hand, the DBAR system defines a transient reference in the x, y, and z axes according to the positional error of the drone and a desired voltage sag. For validation, the proposed controller’s performance is compared with a proportional–integral–velocity (PIV) controller from the literature in five real-time scenarios. The experimental results demonstrate that the proposed controller outperforms the PIV controller in terms of trajectory tracking, regulatory control, and flight time extension. Furthermore, the proposed DBAR system allows the MPC to substantially extend the flight duration without significantly affecting regulatory control and trajectory tracking performance, henceforth, unexpected premature system shutdowns are prevented.