Self-tuning of two-axis autopilot for small aeroplanes and aerial robots

Abstract

A novel, quick and easy-to-implement method of self-tuning of autopilot parameters is presented. Based on oscillatory system excitation and measurement of characteristic parameters which serve as input to control loop parameter computation, the method proposed is targeted at consumer and UAV autopilot implementation, where the control-theory knowledge of operator as well as the a-priori knowledge about the system is at a minimum level. The paper presents both phases of the proposed method, the excitation phase and its automation to maintain safe and bounded signal levels, which are manifested in controllable flight with no departures from the operational envelope in flight, followed by the computatinal phase, where control loop parameters are determined. Simulation results for linear and non-linear models are given, together with considerations and improvements for future work.