An Application to Low‐level Control of Quadrotors
Jun Liu, Milad Farsi · 2022
This chapter presents a case study on the structured online learning (SOL)-based control of quadrotors. For an effective fly of the quadrotor, there are two levels of controls involved: the low-level control that is required for the stability of the hovering position, and the high-level control that provides a sequence of setpoints as commands to achieve a particular objective. To improve the runtime results of the learning, the chapter implements SOL with the recursive least squares algorithm that is more suitable for online applications. The simulation results, illustrates rapid and efficient learning, where an initial model is obtained by random preruns and then the model was improved within different runs in a closed-loop form. Based on the flight data and runtime results, the approach can be employed to automate the control of the quadrotor.