Development of Tactile Imaging for Underwater Structural Damage Detection.

Area of Science:

• Robotics
• Marine Engineering
• Computer Vision

Background:

• Underwater structural inspection traditionally relies on diver visual assessments.
• Remotely operated vehicles (ROVs) offer improved efficiency but lack full autonomy.
• A need exists for autonomous robotic systems for efficient and reliable underwater structural integrity monitoring.

Purpose of the Study:

• To design, prototype, and test a novel Tactile Imaging System for Underwater Inspection (TISUE).
• To enable autonomous and robotic underwater structural damage inspection.
• To evaluate the system's performance in capturing high-quality tactile images in challenging environments.

Main Methods:

• Development of a TISUE system with distinct imaging and manipulation subsystems.
• Novel imaging subsystem utilizing an elastomer-enabled contact-based optical sensor and artificial lighting.
• Laboratory testing of the integrated TISUE system, including data processing and analytics.
• Implementation and training of a deep learning framework for damage detection on tactile images.

Main Results:

• The TISUE system successfully acquired high-resolution, high-quality tactile images of structural surface damage.
• Effective imaging was achieved even in a turbid water environment, demonstrating robustness.
• The deep learning model showed comparable detectability for five damage types between tactile and land-based images.

Conclusions:

• The developed TISUE system provides a viable solution for autonomous underwater structural inspection.
• Tactile imaging is effective for detecting structural damage in challenging underwater conditions.
• The integration of tactile imaging with deep learning offers a promising approach for future underwater structural health monitoring.