Design and implementation of PIλDμ controller for ROVs: Thruster modeling, controller parameter optimization, and FPGA realization

Abstract

Remotely operated and autonomous underwater vehicles (ROVs/AUVs) operate in a harsh environment dominated by nonlinear hydrodynamics, strong coupling, and wave-current disturbances. In most of the existing literature, surge-axis motion is still regulated by integer-order PID controllers that are tuned either heuristically or via single-scenario optimization. Such designs often exhibit limited robustness: their performance degrades significantly under severe noise, targeted wave excitation, or time-varying operational profiles. These limitations motivate the use of fractional-order control and more systematic tuning procedures. This paper investigates fractional-order (PID mu)-D-lambda (FOPID) controllers for surge control and compares two popular meta-heuristics, Particle Swarm Optimization (PSO) and Differential Evolution Algorithm (DEA), in comparison with classical PID. A fourth-order surge plant model is first obtained via system identification of experimental data from a BlueRobotics T200 thruster. Then, PSO and DEA are used to tune both PID and (PID mu)-D-lambda parameters over a multi-scenario cost function that combines step-response quality, disturbance rejection, and control effort. The resulting controllers are evaluated under four increasingly demanding tests: noiseless step tracking, severe white-noise excitation, sinusoidal storm disturbance, and a final scenario with time-varying set-points under the same storm condition. Across all 16 scalar performance metrics (IAE, ISE, and, ITAE over four tests), the DEA-tuned (PID mu)-D-lambda achieves the best value in 12 cases, consistently outperforming both PID designs and the PSO-based (PID mu)-D-lambda. In the most demanding final test (multi-level reference + storm), it reduces the integral time-weighted absolute error ITAE from 0.1065 (best PID) to 0.0893, i.e., by approximately 16%, while preserving competitive control effort. These results provide quantitative evidence that DEA-tuned (PID mu)-D-lambda offers a more robust and energy-aware solution for single-axis surge control in ROV/AUV applications.