A Novel Water-Shore-Line Detection Method for USV Autonomous Navigation.

Area of Science:

• Robotics and Autonomous Systems
• Computer Vision
• Environmental Monitoring

Background:

• Accurate water-shore line (WSL) detection is crucial for unmanned surface vehicle (USV) navigation.
• Existing methods often struggle with non-straight WSL common in inland waterways.
• Boat-borne vision systems require robust WSL recognition for intelligent operation.

Purpose of the Study:

• To develop a novel, adaptable method for detecting both straight and curved water-shore lines (WSL).
• To enhance the intellectualization and navigation capabilities of unmanned surface vehicles (USVs) in complex inland environments.
• To overcome limitations of current WSL detection techniques that primarily focus on straight lines.

Main Methods:

• A three-step approach utilizing image sequences from boat-borne vision.
• Initial line segment pool generation using the Line Segment Detector (LSD) algorithm.
• Coarse-to-fine strategy incorporating water area instability and epipolar constraints for onshore line segment selection.
• Generation of complete shore areas from multi-frame onshore line segments to identify the WSL.

Main Results:

• The proposed method successfully detects both straight and curved water-shore lines (WSL) in inland river scenes.
• Field experiments demonstrate superior adaptability and accuracy compared to existing image processing-based detection algorithms.
• The method effectively generates complete shore areas, with the lower boundary defining the WSL.

Conclusions:

• The novel three-step WSL detection method offers enhanced accuracy and robustness for USV navigation.
• This approach significantly improves the ability to navigate complex inland waterways with variable shorelines.
• The findings contribute to the advancement of intelligent navigation systems for unmanned surface vehicles.