Robust depth–yaw control of autonomous underwater vehicles under disturbances and parametric uncertainties

This paper presents a simulation-based comparative investigation of four control strategies for depth and yaw regulation of a nonlinear six-degree-of-freedom (6-DOF) Autonomous Underwater Vehicle (AUV) operating under external disturbances and parametric uncertainties. Proportional–Integral–Derivative (PID) control, Fuzzy Logic Control (FLC), Super-Twisting Sliding Mode Control (ST-SMC), and Nonlinear Model Predictive Control (NMPC) are implemented within a unified MATLAB/Simulink framework and evaluated under three operating scenarios: nominal conditions, external ocean-current disturbances, and hydrodynamic parameter uncertainties. Controller parameters are held fixed across all scenarios to ensure a consistent comparison. Performance is assessed in terms of trajectory-tracking accuracy, root-mean-square error (RMSE), robustness, control effort, and computational execution time. The results show that ST-SMC provides the most consistent tracking performance among the four controllers. For depth regulation, ST-SMC achieves RMSE values of 0.0920 × 10⁻², 0.9903 × 10⁻², and 0.1819 × 10⁻² m under nominal, disturbed, and uncertain conditions, respectively, with corresponding yaw RMSE values of 0.0250 × 10⁻², 0.3276 × 10⁻², and 0.0270 × 10⁻² rad. These results indicate improved tracking accuracy and robustness relative to PID, FLC, and NMPC across the scenarios considered. Real-time computational performance was profiled at the common 1 ms sampling period used throughout the simulation campaign. PID exhibits the lowest computational cost, with an average execution time of 0.0011 ms per call. ST-SMC also shows very low computational demand, with an average execution time of 0.0142 ms, satisfying the 1 ms deadline in 99.903% of the 100,001 profiled samples. In contrast, FLC and NMPC require substantially higher computational effort, with average execution times of 11.37 ms and 11.0057 ms, respectively, exceeding the 1 ms deadline in all profiled samples; their practical deployment therefore presupposes slower control-update rates or more capable embedded hardware. All four controllers achieved a 100% feasible-solution rate throughout the simulation campaign. Overall, ST-SMC offers the most favorable balance between tracking accuracy, robustness, and computational cost among the investigated strategies, providing a consistent benchmark of the four control approaches within a unified simulation framework. As the reported computational measurements were obtained in a Software-in-the-Loop environment, further Hardware-in-the-Loop and experimental validation on physical AUV platforms is required to confirm real-time performance under operational conditions.