Position and Orientation Control of Remotely Operated Vehicles Utilizing Radial Basis Function Neural Networks to Overcome Underwater Environmental Disturbances

Conventional control systems, such as Proportional Integral Derivative (PID), have difficulty adapting to changes, which reduces the stability and efficiency of Remotely Operated Vehicles (ROVs). This study develops an adaptive control system based on Radial Basis Function Neural Network (RBFNN) combined with PID to improve PID adjustment and ROV stability. The research methodology involves designing RBFNN and PID controls, simulating them in MATLAB/Simulink, and testing them in three scenarios: calm currents, strong currents, and random disturbances. The uniqueness of this study lies in how it overcomes existing challenges and offers a new approach to ROV control in dynamic environments by comparing 6- and 8-thruster configurations on ROVs. The simulation results show that the application of RBFNN successfully reduces oscillations in the X position (0–2.5 m), Y position (-1 m), Z position (vertical drift), Roll, Pitch, and Yaw orientations. After RBFNN control, the ROV's position and orientation become more stable and closer to the setpoint, with an error of less than 1%. This study shows that MK-RBFNN-based adaptive control with an 8-thruster configuration can improve the stability and responsiveness of ROVs in dynamic environmental conditions. The significance of this research lies in the challenge of controlling ROVs for underwater structure inspection in dynamic environmental conditions.