Comparative analysis of data-driven predictive control methods for quadrotor systems

Date

2025-3-20

Author

Gedik, Ramazan Kürşat

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

90
views

0
downloads

Cite This

Traditional control methods are widely employed in the industry and are mostly based on mathematical models representing system dynamics. While effective, these mathematical models may fail to capture the full complexity of real-world systems due to uncertainties, nonlinearities, or unmodeled dynamics. On the other hand, data-driven methods provide an alternative to model-based methods, using empirical data instead of relying exclusively upon a mathematical model. These methods employ some form of optimization algorithm in order to connect real-time measurements to previously collected system data. The controller designed based on a data-driven approach can operate without having the system model in detail. This study explores the implementation of data-driven predictive controllers on a quadrotor model to achieve trajectory tracking. Three different data-driven methods are evaluated regarding tracking performance and computational efficiency. The study aims to determine the most suitable data-driven control approach for quadrotor trajectory tracking, considering trade-offs between accuracy and computing time. Additionally, the best-performing data-driven method is compared with a conventional linear Model Predictive Controller (MPC) under conditions of parameter uncertainty in the quadrotor model. The comparison proves that data-driven methods are robust enough against uncertainties in the system dynamics. This study highlights the significant potential of data-driven predictive control for managing large-scale systems with uncertainties, comparing it with model-based controllers. The findings contribute meaningfully to the development of advanced control strategies, particularly in the field of aerial robotics.

Subject Keywords

Data-driven, Control, Tracking, Robotics

URI

https://hdl.handle.net/11511/114165

Collections

Graduate School of Natural and Applied Sciences, Thesis