Robust Learning Control for Shipborne Manipulator With Fuzzy Neural Network.

Area of Science:

• Robotics
• Marine Engineering
• Control Systems

Background:

• Shipborne manipulators are crucial for marine vehicle autonomy.
• Conventional proportional-derivative (PD) controllers struggle with accuracy and safety in dynamic ocean environments.
• Existing control methods often require precise tuning and accurate manipulator dynamics models.

Purpose of the Study:

• To develop a real-time, adaptive control approach for shipborne manipulators.
• To improve the position control performance of manipulators under ocean conditions.
• To overcome the limitations of conventional PD controllers in uncertain environments.

Main Methods:

• Implementation of a parallel PD controller and a fuzzy neural network (FNN) for PD+FNN control.
• Integration of a sliding mode control (SMC) theory-based learning algorithm.
• Validation through qualitative and quantitative simulations and real-world experiments.

Main Results:

• The PD+FNN controller demonstrated superior performance compared to the conventional PD controller.
• The approach effectively handles uncertainty and disturbance in ocean conditions.
• Angle compensation deviation was improved, achieving results within ±1°.

Conclusions:

• The proposed PD+FNN control strategy offers a robust and adaptive solution for shipborne manipulators.
• This method reduces the need for precise controller tuning and accurate dynamic models.
• The adaptive controller ensures safe and accurate operations for marine autonomous systems.