A Cross Structured Light Sensor and Stripe Segmentation Method for Visual Tracking of a Wall Climbing Robot

Area of Science:

• Robotics and Automation
• Non-Destructive Testing
• Computer Vision

Background:

• Outdoor non-destructive testing (NDT) of metal welds faces challenges from ambient light noise affecting structured light sources.
• Accurate weld line tracking is crucial for automated inspection and robot navigation in industrial settings.

Purpose of the Study:

• To develop a robust system for weld line detection and tracking in challenging outdoor environments.
• To improve the accuracy and reliability of non-destructive testing for metal welds.

Main Methods:

• Design of a cross structured light (CSL) sensor for weld line detection.
• Development of a robust laser stripe segmentation algorithm using adaptive monochromatic space and minimum entropy deconvolution.
• Integration of the CSL sensor and algorithm for guiding a wall climbing robot.

Main Results:

• The CSL sensor accurately captures 3D weld information, even with environmental noise.
• The proposed segmentation algorithm effectively overcomes noise, enabling reliable stripe center extraction.
• Successful application in guiding a wall climbing robot for wind power tower weld inspection.

Conclusions:

• The developed CSL sensor and algorithm provide a robust solution for outdoor weld line inspection.
• The system enhances the accuracy of 3D weld information capture and robot navigation for NDT applications.