Closed-loop Indirect Identification of a Quadrotor Generalized Attitude Model Using Outdoor Flight Data

In terms of outdoor flight experimental data, the paper investigates the indirect identification of a generalized attitude model (GAM) for a controlled quadrotor. The GAM is comprised of an attitude model of the quadrotor and the dynamics of the actuator involved in the attitude control loop. Being collected from outdoor, the flight data are undoubtedly corrupted by large-magnitude random wind and other unknown disturbances. Another challenge is from an existing unknown attitude controller. In such case, an indirect identification scheme is proposed to estimate the quadrotor GAM. The correlation analysis is first applied to reduce the disturbance impacts in the flight data as greatly as possible. A pair of non-coprime factors of the GAM can then be estimated from the available correlation processing experimental data. An extended Hankel model reduction method is studied to cancel the same unstable zeros and poles in the non-coprime factors for obtaining their corresponding coprime factor pair, from which the GAM is readily estimated. Finally, model validations are given to show the accuracy of the estimated GAM.