Anti-Swing Trajectory Tracking of Slung-Payload UAVs: An Attention-Infused Neural Observer Approach With Output Constraints

Slung-payload quadrotor uncrewed aerial vehicles (UAVs) are pivotal in aerial logistics but suffer from performance degradation due to payload swing, strong coupling dynamics, and external disturbances. To address these challenges, this article proposes a constrained adaptive control scheme integrated with an attention-based neural observer. First, an enhanced cascade framework explicitly captures the UAV-payload coupling to generate active anti-swing commands. To ensure operational safety, barrier Lyapunov functions (BLFs) are employed to strictly confine position errors within prescribed safety constraints, preventing constraint violation even during complex trajectory tracking. Furthermore, a novel disturbance observer leveraging an attention mechanism is developed. Unlike conventional estimators, this mechanism selectively weighs historical error features to accurately reconstruct time-varying aerodynamic disturbances. Rigorous stability analysis proves that all closed-loop signals are uniformly ultimately bounded. Finally, comparative simulations and real-world flight experiments demonstrate the superior tracking accuracy and swing suppression of the proposed strategy.