Relay-based identification of Aerodynamic and Delay Sensor Dynamics with applications for Unmanned Aerial Vehicles

TL;DR

This paper introduces a real-time, relay feedback-based system identification method for UAVs that accurately estimates aerodynamic and sensor delay parameters onboard, reducing reliance on costly lab tests and adapting during operation.

Contribution

It presents a novel real-time identification approach using MRFT and LPRS, enabling onboard dynamic parameter estimation for UAVs with high accuracy and speed.

Findings

Method accurately estimates UAV aerodynamic parameters.

Identification process takes only a few seconds.

Outperforms existing methods in accuracy and uncertainty quantification.

Abstract

In this paper, we present a real-time system identification method based on relay feedback testing with applications to multirotor unmanned aerial vehicles. The proposed identification method provides an alternative to the expensive lab testing of certain UAV dynamic parameters. Moreover, it has the advantage of identifying the parameters that get changed throughout the operation of the UAV, which requires onboard identification methods. The modified relay feedback test (MRFT) is used to generate stable limit cycles at frequency points that reveal the underlying UAV dynamics. The locus of the perturbed relay system (LPRS) is used to predict the exact amplitude and frequency of these limit cycles. Real-time identification is achieved by using the homogeneity properties of the MRFT and the LPRS which are proven in this paper. The proposed identification method was tested experimentally to estimate the aerodynamic parameters as well as the onboard sensor's time delay parameters. The MRFT testing takes a few seconds to perform, and the identification computations take an average of 0.2 seconds to complete in modern embedded computers. The proposed identification method is compared against state-of-the-art alternatives. Advantages in identification accuracy and quantification of uncertainty in estimated parameters are shown.