What Is T2D2? The Thornton Tomasetti Damage Detector is a technology we developed to automatically detect visible damage in various types of structures: • Concrete (cracks, spalls, exposed rebar) • Steel (corrosion, pitting, tears) • Masonry (cracks, spalls, efflorescence) • Timber
We use engineering principles to solve the world’s challenges – starting with yours. From practical tasks to creative asks, we have your solution. We are a team of inventive problem solvers who understand
How Does It Work?
that collaboration leads to better results. When faced with any challenge, we look at the big picture
The application uses the latest developments in artificial intelligence, specifically deep learning for computer vision, to detect and localize damage by analyzing inspection images. It has been trained on labeled data sets consisting of innumerable images collected through years of manual inspections on various structures with different damage classes.
and work to nail the details while fulfilling your broader goals.
We want to hear from you! Badri Hiriyur can be reached at BHiriyur@ThorntonTomasetti.com.
Sept em ber 2019
Offices Worldwide E-mail us if you would like to collaborate or learn more about our services. We would love to stop by your office and learn about your project aspirations and requirements.
T2D2
51 Madison Avenue New York, NY 10010 T +1.917.661.7800
ThorntonTomasetti.com
An AI Revolution in Structural Health Assessment
Image Acquisition
rebar: 65% spall: 37%
crack: 48%
Photogrammetry rebar: 65%
crack: 40% crack: 39%
Damage Detection
Assessment Reporting The T2D2 mobile app’s onboard AI inference engine performs real-time detection and reports its confidence that an observation falls into a particular damage class.
Drone Deployment
Mobile/Camera Deployment
T2D2 can be deployed on a drone for image capture at scale or to reach inaccessible spaces without special equipment or scaffolds.
The same underlying technology can be used manually as a mobile app to capture images and detect damage (and localize via GPS) in real time as an inspector walks around a site.
zoomteam/Евгений Косцов /undrey/ Sergei Poli vanov © 123rf.com
Automated detection of cracks, spalls and other material failures
spall: 33%