System Identification for Small, Low-Cost, Fixed-Wing Unmanned Aircraft

Abstract

This paper describes a practical system identification procedure for small, low-cost, fixed-wing unmanned aircraft. Physical size and cost restrictions limit the sensing capabilities of these vehicles. The procedure is demonstrated on an Ultra Stick 25e, therefore emphasizing a minimum complexity approach compatible with a low-cost inertial sensor. A linear model, obtained from the generic nonlinear equations of motion for aircraft, is used as a basis for system identification. This model is populated with results from a first principles analysis to form a baseline model. Flight experiments are designed using the baseline model and operational constraints to collect informative data. Parameters of the linear model are identified by fitting the model to frequency responses extracted from the data. The parameters are integrated into the nonlinear equations of motion, and both linear and nonlinear models are validated in the time domain. Verification of model accuracy is extended with a sensitivity and residual analysis.