RABIT™ (Robotics-Assisted Bridge Inspection Tool) is an autonomous bridge assessment tool designed and developed as part of the Long-Term Bridge Performance (LTBP) Program by Federal Highway Administration, Center for Advanced Infrastructure and Transportation and Rutgers School of Engineering. RABIT collects data using several cutting-edge nondestructive evaluation (NDE) technologies and provides comprehensive condition assessment of concrete bridge decks. The NDE sensors like ground-penetrating radar, ultrasonic surface waves and high resolution cameras detect defects like corrosion, delamination and cracks in concrete, without harming the bridge structure. In addition, a panoramic camera images the surroundings. Manually collecting all the data is time-consuming, labor intensive, limits the amount of data that can be collected and causes traffic disruptions. RABIT eases the process of data collection and analysis, making it quick and precise and provides an overall assessment of the bridge decks using multiple sensors. This project was awarded the 2014 Charles Pankow Award for Innovation by the American Society of Civil Engineers (ASCE).
I was part of the computer vision team supervised by Dr. Kristin J. Dana. I worked on the Ground Penetrating Radar (GPR) sensor, which captures information from the subsurface of concrete bridges and represent it as high quality images. Reinforced concrete (RC) bridges have steel reinforcement bars (rebars) embedded within the decks for structural strength and they form a distinct hyperbolic signature in the images. Rebars deteriorate over time and their signature in images varies due to decreased signal strength. We developed methods to interpret the GPR images with pattern recognition and machine learning to find the rebar hyperbolic signature. Our approach used image-based gradient features and robust curve fitting. The detected hyperbolic signatures of rebars within the bridge deck are used to generate deterioration maps of the bridge deck. We compared the results of the rebar region detector quantitatively with several methods of image-based classification and demonstrated a significant performance advantage of using the new method. The traditional methods for obtaining deterioration maps from GPR data often require manual interaction and offsite processing. Application of this new algorithm along with robot integration, automates the process of bridge deck assessment.
More Information: RABIT™ Bridge Deck Assessment Tool