Bearing Rigidity-Based Flocking Control of AUVs via Semi-Supervised Incremental Broad Learning.

Area of Science:

• Robotics
• Marine Engineering
• Control Systems

Background:

• Flocking control is crucial for autonomous underwater vehicle (AUV) coordination in marine missions.
• Existing methods face challenges in weak communication and complex environments.
• A research gap exists in robust flocking strategies for AUVs under these conditions.

Purpose of the Study:

• To address the research gap in AUV flocking control within weak communication and complex marine environments.
• To develop a robust and efficient flocking strategy for AUVs.
• To enhance localization accuracy and energy efficiency in AUV coordination.

Main Methods:

• Utilized graph theory, specifically bearing rigidity graphs, to model AUV topology.
• Developed an iterative gradient descent-based localization estimator.
• Implemented a min-weighted bearing rigidity graph strategy for improved localization and energy efficiency.
• Employed a semi-supervised broad learning system (BLS) for model-free flocking controllers in obstacle-rich environments.

Main Results:

• The min-weighted bearing rigidity graph strategy balances localization accuracy and communication consumption effectively.
• The semi-supervised BLS controller reduces training time and overcomes label limitations compared to supervised methods.
• Simulation and experimental studies validated the proposed flocking control strategies.

Conclusions:

• The proposed min-weighted bearing rigidity graph localization strategy offers a superior balance between accuracy and communication costs.
• The semi-supervised BLS-based flocking controller provides an efficient solution for AUVs in complex environments with limited labeled data.
• The integrated approach enhances the feasibility of sophisticated marine coordination missions for AUVs.