article · IEEE Access
Controlling the flight paths of quadrotor unmanned aerial vehicles usually relies on methods like proportional-integral-derivative controllers or fuzzy versions of them. These traditional controllers require extensive tuning and precise knowledge of complex flight dynamics. A new control framework combines fuzzy logic, conventional control principles, and recurrent neural networks to improve trajectory tracking without requiring explicit dynamic models. The recurrent neural network learns to reproduce the control responses of a fuzzy proportional-integral-derivative controller, uniting fuzzy reasoning with data-driven neural adaptability. Comparative evaluations conducted through comprehensive numerical simulations show that this combined controller outperforms both standard and fuzzy controllers across various performance measures. It delivers superior tracking precision, greater robustness, and enhanced adaptability when handling complex, nonlinear flight behaviours.
Unmanned aerial vehicles must navigate unpredictable environments accurately and reliably. By using neural networks to reduce the reliance on complex mathematical models and manual tuning, this method simplifies the implementation of flight control systems. This supports more dependable autonomous flight, better trajectory tracking, and improved resilience against sudden disruptions during operation.
This research is an early-stage study focused on flight control software for quadrotor unmanned aerial vehicles. The method could enable drone manufacturers and navigation software developers to achieve higher trajectory accuracy without complex dynamic modelling. Because the findings are based entirely on numerical simulations, the technology requires physical flight testing before it can be integrated into commercial products.
AI-generated from the published abstract. Always read the original work before citing.
This paper presents a novel Fuzzy PID-based Recurrent Neural Network (FPIDRNN) controller designed to enhance trajectory control in quadrotor Unmanned Aerial Vehicles (UAVs). Conventional control approaches such as Proportional-Integral-Derivative (PID) and Fuzzy PID (FPID) require extensive tuning and a deep understanding of system dynamics. On the other hand, the FPIDRNN controller provides a data-driven alternative that adapts effectively to complex, nonlinear behaviors without the need for explicit dynamic modeling. By training the neural network to replicate the control actions of the Fuzzy PID controller, this approach combines the strengths of fuzzy logic and neural adaptation, resulting in robust and precise control. A comparative analysis of three controllers—PID, Fuzzy PID (FPID), and the proposed FPIDRNN—is conducted in the context of quadrotor UAV control. Extensive numerical simulations demonstrate that the FPIDRNN controller significantly enhances tracking accuracy, robustness, and adaptability compared to both the PID and FPID controllers, as evidenced by various performance indices. This paper underscores the efficiency of the FPIDRNN as an advanced control solution, integrating fuzzy logic, PID, and neural network techniques to elevate UAV control systems.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1109/access.2024.3516494
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
