System identification driven high-performance controller design for an agile small rotorcraft UAV

Abstract

This paper presents an efficient methodology for improving an existing legacy controller for the MIDIARTIS, a very agile 10 kg rotorcraft Unmanned Aerial Vehicle (UAV) by leveraging the same legacy controller as a basis for system identification flight tests. The legacy autopilot was modified to enable the injection of system identification signals into the control reference values and to actuator signals during flight. This allowed the excitation of low, middle and high frequencies of the rotorcraft UAV with an active controller stabilizing the UAV either in hover condition or on a trajectory during forward flight. The resulting data was then used to identify a physical 9-degrees of freedom model, accounting for rotor flapping and engine states. Based on this model, an improved flight controller was developed which is shown to greatly improve agility and precision of the rotorcraft UAV.