Railway Interchange 2019

Using 3D Laser and Artificial Intelligence to Inspect Wooden and Concrete Ties (Room 103 A)

24 Sep 19
8:30 AM - 9:00 AM

Tracks: AREMA Technical Sessions By Date- Maintenance-of-Way, AREMA Technical Sessions By Functional Group- Maintenance-of-Way, Technical Sessions By Day- Tuesday

Railway networks around the world represent billions of dollars of investment both in terms of their initial construction as well as their ongoing maintenance. Poorly maintained networks pose a risk to asset life, schedule and safety. This presentation will describe a new approach which leverages 3D laser triangulation as an imaging solution and computer-based Deep Learning as an inspection method to inventory and inspect the condition of wooden and concrete ties. 3D laser triangulation captures a both a high-resolution image (2D) and a 3D point cloud of the entire track area and can be used at revenue speeds, day or night. Deep Learning is a type of machine learning wherein the computer learns to solve a complex problem in a way that is similar to the way in which humans learn (using a neural network). Deep Learning is well-suited to the task of image analysis and has even outperformed humans in tasks such as the detection of cancer cells in the medical field. This presentation will include an overview of the principle of operation of 3D laser triangulation sensors as well as Deep Learning. Sensor scanning frequency, resolution and accuracy will also be discussed. The Deep Learning algorithm training process will be discussed and numerous examples of AI-based inspections of both wooden and concrete ties will be presented. Attendees will leave with a good overview of the principle of operation, resolution and accuracy, state of development, and potential application areas of this exciting new technology for their network. *Missed this Session or want to view it again? Stop by the VirtualAREMA kiosk located in the Registration area to purchase this product today!