Robust Underwater Docking Visual Guidance and Positioning Method Based on a Cage-Type Dual-Layer Guiding Light Array.

Area of Science:

• Robotics
• Computer Vision
• Marine Engineering

Background:

• Autonomous Underwater Vehicles (AUVs) require precise localization for docking.
• Limited camera fields of view cause guiding beacons to be lost during AUV approach.
• Existing methods struggle with intermittent beacon visibility and varying appearances.

Purpose of the Study:

• To develop a robust visual localization method for AUV docking.
• To overcome challenges of beacon visibility loss and intermittent data.
• To enable continuous visual guidance for autonomous AUV recovery.

Main Methods:

• A cage-type, dual-layer guiding light array system is proposed.
• A dual-layer light array localization algorithm dynamically distinguishes between front and rear arrays.
• Particle Swarm Optimization (PSO) and Backtracking Extended Kalman Filter (BTEKF) are used for tag matching and error correction.

Main Results:

• Achieved high matching success rates (90.3% with 3 missing lights, 99.6% with 2 spurious beacons).
• Error correction with BTEKF improved success rate to 99.9%.
• Demonstrated stable and robust tag matching with an average detection time of 0.112 s.

Conclusions:

• The proposed dual-layer guiding light array system provides reliable visual guidance for AUV docking.
• The method ensures continuous localization and robust tag matching under challenging conditions.
• Enables stable and continuous visual guidance-based docking for autonomous AUV recovery.