Dynamic battery-aware auxiliary reference system for improving real-time model predictive control implementation in quadrotors

To perform better trajectory tracking, numerous advanced control strategies have been implemented in quadrotors. However, drone efficiency in practical tasks is tied to its flight autonomy, and algorithms do not take into account the energy consumption rate. Moreover, even though battery management systems have been designed to prevent overdischarges, they do not consider voltage sags caused by thrust variations. In this work, a dynamic battery-aware auxiliary reference system is presented. First, the relationship between the drone's positional error and the discharge rate due to the control action is modeled. Then, according to the actual drone's positional error and the desired voltage sag, a transient reference in the z-axis is defined. For validation, three scenarios are conducted in real-time. Thus, the performance of the control strategy described in previous work is compared with the performance of the model predictive control strategy using the proposed system. The experimental results show that, when the controller uses the proposed system, the flight time is extended, without notably affecting the trajectory tracking performance and avoiding unexpected system shutdowns.