Neuro-Fuzzy Dynamic Position Prediction for Autonomous Work-Class ROV Docking.

Area of Science:

• Robotics and Autonomous Systems
• Ocean Engineering
• Artificial Intelligence

Background:

• Remotely Operated Vehicles (ROVs) face challenges matching surface vessel heave motion due to power, drag, and inertia limitations.
• Current ROV docking relies on manual pilot control, which is unsuitable for autonomous operations.
• Accurate prediction of docking station heave motion is crucial for dynamic underwater docking maneuvers.

Purpose of the Study:

• To develop and present an Adaptive Neuro-Fuzzy Inference System (ANFIS)-based method for predicting docking station heave motion.
• To enable more reliable and accurate autonomous docking for work-class ROVs.
• To provide a solution for the limitations of human-in-the-loop control in dynamic ROV docking.

Main Methods:

• Implementation of an Adaptive Neuro-Fuzzy Inference System (ANFIS) for heave motion prediction.
• Utilizing real-world trajectory data recorded during offshore trials in the North Atlantic Ocean.
• Testing the ANFIS model with data from a work-class ROV and a cage type Tether Management System (TMS).

Main Results:

• The ANFIS-based method demonstrated effective prediction of docking station heave motion.
• The system's performance was validated using actual offshore trial data.
• The proposed method shows promise for enhancing autonomous docking capabilities.

Conclusions:

• The ANFIS approach offers a viable solution for predicting docking station heave motion, crucial for autonomous ROV operations.
• This predictive capability can significantly improve the safety and efficiency of dynamic underwater docking.
• The study highlights the potential of AI-driven methods in subsea robotics and autonomous systems.