Abstract

Accurate modelling of a quadrotor is essential for control design and stability analysis because its dynamics vary significantly across different operating conditions. In this investigation, data-driven system identification is performed around several operating points, including hover, forward flight, and lateral/sideways flight. A pseudo-random binary sequence (PRBS) excitation generated output data in the presence of a proportional-integral-derivative (PID) controller. Three different identification methods are applied, and their results are compared: the prediction error (PEM), least-squares auto-regressive with exogenous input (ARX), and output error (OE) methods. To verify identification accuracy, the fit percentage is recorded for all three methods. Furthermore, pole analysis and step-response evaluations are conducted to assess performance. The results show that the PEM based transfer function identification method provides the highest fit percentage across different operating points compared to the other two methods. To validate the suitability of the identified model for control design, the model is subsequently stabilized using a PID controller.

About the Speaker

Ms. Bushra Shaikh is a Graduate Student in the Control and Instrumentation Engineering Department at King Fahd University of Petroleum & Minerals (KFUPM), Saudi Arabia. She also served as a Lecturer at Mehran University of Engineering & Technology (MUET), Pakistan. Her research interests include quadcopter modeling, system identification, and control design, as well as multi-agent coordination and cooperative autonomy for networked robotic systems. She aims to develop reliable and scalable control and estimation strategies that can be effectively applied to real-world engineering systems.