Grey-box Real-time Identification of Constrained-parameter UAV Systems based on Relay-based Limit Cycle Oscillations

Abstract

Recent advances in the development of unmanned aerial vehicles (UAVs) have unlocked their potential for applications where safety and reliability are fundamental, particularly in presence of unforeseen uncertainties. Reliable UAV dynamics identification is necessary to enable real-time adaptation to any changes in flight conditions, through fault detection and state estimation. In this work, a grey-box real-time identification of UAV systems is presented. The approach enables real-time identification of various system parameters such as the drag coefficient, communication and processing time delay, and the force provided by the propulsion system. The real-time performance is achieved using a novel bounded identification space, referred to as Unit Frequency Manifold (UFM), which leverages homogeneous properties of the modified relay-feedback test (MRFT). During the test, the system under investigation is shown to maintain stability which is required for testing open-loop unstable systems such as a UAV. The method is tested in simulation and is shown to achieve higher accuracy compared to the describing function method available in the literature. This is attributed to the exact nature of the proposed identification method. The identification method developed in this work was used to identify altitude, attitude and lateral dynamics of a UAV, the design of which adheres to the constraints provided in the theoretical description of this work. As an application of the real-time system identification algorithm, a human interaction detection algorithm that generalizes to different UAVs was developed based on the identified parameters.

Date of Award	Aug 2023
Original language	American English
Supervisor	Yahya Zweiri (Supervisor)