Longitudinal System Identification of Ornithopter with Automated Flight Tests**This research was supported by a grant to Bio-Mimetic Robot Research Center funded by Defense Acquisition Program Administration (UD130070ID).

Abstract

The purpose of this paper is to identify a longitudinal linear model of an ornithopter by automated flight tests. Flight tests were conducted outdoors, and an avionic board with sensors was installed on-board for measuring angular rates, Euler angles, and a total velocity. For accurate flight test, automated signal input is designed for elevator deflection: doublet and multi-step 3211 maneuver. During a cruise flight, the ornithopter normally had oscillation which is generated by flapping motion of the main wings. Fast Fourier transform (FFT) is used for analyzing flight data in a frequency domain, and a Butterworth filter is designed to filter the corrupted data by the flapping motion. The structure of the ornithopter linear model is assumed to be similar to a fixed-wing aircraft which has a periodic oscillation because it has similar control surfaces except the flapping motions. For system identification, unknown parameters are estimated by unconstrained nonlinear optimization.