Computationally simple model reference adaptive control for miniature air vehicles

Abstract

This paper presents a computationally efficient model reference adaptive autopilot for pitch attitude hold and roll attitude hold. The derivation of the adaptive controllers depends upon a model simplification that results in two unknown parameters for each mode. The resulting controllers have been successfully implemented on the Kestrel autopilot (which has a low-power micro-controller) and flight tested on three different miniature air vehicles (MAVs). In addition, one of the MAVs was outfitted with a deployable flap that significantly changes the aerodynamic coefficients of the airframe. The adaptive controllers proposed in this paper successfully adapt to mid-air deployment of the flap.