3D Vision-Guided Adaptive 3D Ultrasonic Scanning for Robotic Arms: Nondestructive Testing of Aerospace Components

Area of Science:

• Materials Science
• Robotics
• Non-Destructive Testing

Background:

• Traditional 3D ultrasonic testing of composite structures is inefficient due to manual operations.
• Existing robotic arms struggle with complex curved surfaces because of fixed trajectory limitations.

Purpose of the Study:

• To develop an automated 3D ultrasonic testing method for aircraft composite laminated structures.
• To address the limitations of manual operation and fixed trajectory planning in current testing methods.

Main Methods:

• A Yolo-Mask model was used for visual recognition and segmentation of composite components.
• 3D vision guidance enabled automatic planning of robotic arm scanning trajectories.
• Point cloud processing and data fusion algorithms were employed for 3D reconstruction.

Main Results:

• The method achieved a Segm-mAP of 97.4% and a detection speed of 11.7 fps.
• 3D imaging error was less than 0.1 mm, demonstrating high precision.
• Fully automated detection was realized, significantly improving efficiency.

Conclusions:

• The proposed vision-guided robotic system offers an effective solution for automated non-destructive testing of aircraft composites.
• This approach enhances detection efficiency and accuracy for complex structures.