Cascaded guidance, state estimation, and control for depth tracking of underactuated autonomous underwater vehicles

Cascaded guidance, state estimation, and control systems have been successfully implemented in autonomous underwater vehicles. However, disordered convergence sequences among subsystems fundamentally induce system oscillations and instabilities. Here, we introduce a temporal-sequencing-convergent cascaded guidance, state estimation, and control system for depth-tracking of underactuated autonomous underwater vehicles. It establishes an ordered convergence architecture: state estimation converges first with a planning window t_d, followed by control execution with a planning time window t_c, and guidance finally converges with a planning time window t_f, adhering to a temporal-sequencing-convergent criterion t_d < t_c < t_f. This architecture, validated through experiments, ensures stable and efficient depth-tracking performance owing to well-ordered convergence of subsystems. Conversely, experiments violating the temporal-sequencing-convergent criterion exhibited prominent oscillatory depth-tracking responses. Our method outperformed selected approaches, achieving a lower average depth-tracking error of 1.32 cm under sudden external disturbances. Notably, even with prominent pitch-tracking errors using a PD controller, our proposed guidance design also showcased remarkable attack-angle compensation capability, thereby maintaining precise depth-tracking performance.