Guidelines for System Identification of Multirotor Vehicles with Highly Correlated Inputs

With the development of new over-actuated multirotor vertical lift aircraft, there is a need to identify accurate vehicle models from flight data at the individual effector level to validate physics based models or skip physics based models altogether and develop simulation models direction from the identification results. Typical frequency-domain system identification methods break down when the bare-airframe inputs are highly correlated, which is often the case when collecting system identification data for over-actuated vehicles. The Joint Input-Output Method has been previously used to mitigate the issue of input correlation, and is applied in this paper to a simulation example based on an octocopter UAS in hover. Several excitation methods are investigated to determine which are best suited for use with the Joint Input-Output Method. The identification results of the different methods with varying levels of measurement noise are presented, as are the main advantage and disadvantaged of each method. Based on these results, guidelines are provided for multirotor system identification using the Joint Input-Output Method.