Automated Transverse Crack Mapping System with Optical Sensors and Big Data Analytics

Area of Science:

• Civil Engineering
• Materials Science
• Computer Vision

Background:

• Transverse cracks on bridge decks are a primary cause of deterioration due to chloride penetration.
• Accurate crack width and spacing data are crucial for assessing bridge health.
• Existing crack detection methods often rely on limited visual characteristics.

Purpose of the Study:

• To develop an automated data pipeline for detecting transverse cracks on bridge decks.
• To create a non-contact optical sensing system for efficient data acquisition without traffic disruption.
• To enhance crack detection by incorporating crack direction and shape analysis.

Main Methods:

• Development of a data acquisition system using non-contact optical sensors.
• Implementation of an alternative localization strategy using only optical sensor data, independent of GPS or odometers.
• Application of a machine learning approach that considers crack direction and shape for improved detection accuracy.

Main Results:

• Successful development of a data pipeline for automated crack detection.
• A novel localization strategy using optical sensors was established.
• The machine learning model demonstrated improved performance by analyzing crack geometry.

Conclusions:

• The proposed system offers an efficient and deployable tool for bridge deck crack inspection.
• The system facilitates big data analytics for monitoring crack progression over time and space.
• This technology can significantly aid in proactive bridge maintenance and management.