Neural Network-Based Hybrid Three-Dimensional Position Control for a Flapping Wing Aerial Vehicle

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Flapping Wing Aerial Vehicles (FWAVs) present unique control challenges due to complex aerodynamics.
• Achieving stable and precise 3-D positioning for FWAVs requires advanced control methodologies.
• Existing control strategies often struggle with dynamic uncertainties and safe operational boundaries.

Purpose of the Study:

• To develop a novel neural network-based hybrid mode-switching control strategy for FWAVs.
• To enhance the stabilization accuracy and flight efficiency of FWAVs in three dimensions.
• To ensure system safety by preventing dangerous flight conditions through adaptive control.

Main Methods:

• A novel dynamic model for FWAVs resolved in a proposed vertical frame.
• A Radial Base Function Neural Network (RBFNN)-based adaptive control strategy with mode-switching.
• Integration of saturation functions and Barrier Lyapunov Functions (BLFs) for velocity constraints.
• Lyapunov techniques and hybrid system analysis for theoretical stability guarantees.

Main Results:

• Successful stabilization of the FWAV to a desired 3-D position.
• Demonstrated reliability and efficiency of the proposed hybrid control strategy through experiments.
• Theoretical guarantee of semiglobal uniform ultimate boundedness for the closed-loop system.
• Effective constraint of lateral velocity within proper ranges.

Conclusions:

• The proposed vertical frame and cooperative switching learning/control strategies represent significant innovations.
• The hybrid mode-switching control strategy offers superior performance for FWAV stabilization.
• Experimental validation confirms the controller's effectiveness and robustness.