Driverless Trains: Not a dream anymore

Written By
Published on April 14th, 2020

AI-powered, driverless trains are coming. New York Air Brake’s driverless train software was used in last summer’s demonstration of a 30-car freight train moving 48 miles at a research and testing facility operated by the American Railroads Association, according to a recent Seattle Times report.

Today, big, heavy freight trains can only operate in Australia, without an onboard engineer. The initiative began when the mining company Rio Tinto started operating driverless trains on its remote rail network to carry iron ore, hundreds of miles. An Australian, Japanese, and US collaboration has been involved in designing the app, called AutoHaul. More than $940 million has been spent to get it up and running.

The AutoHaul system processes data about the location, length, and weight of the train to determine the ideal speed to safely move the train.

Railroad companies also use AI machine learning in a variety of ways outside of driverless trains.

Texas railroad company BNSF uses AI to enhance the way it handles its freight and track vehicles. For example, the company uses computer vision to analyze 750,000 images every day for the identification of broken wheels.
The groundwork for remote train operations is being laid with the introduction of Positive Train Control, a program designed to avoid train collisions and speed-related derailments, suggested David Clarke, a professor at the University of Tennessee and director of the Transportation Research Center at the school

AI’s advantages in the railway industry include managing and synchronizing timetables to minimize overall travel time, waiting time and transferring waiting time, outcomes that improve customer loyalty and can also save costs.

GE Movement Planner forecasts train traffic trends and seeks to improve the velocity, capacity, and performance of railroads. Wabtec Corp. has acquired GE Movement Planning in a deal completed in February 2019. According to a Wabtec press release, revenue from GE Movement Planner goods was projected at about $1 billion in 2019. The value of the deal was $11.1billion.

A similar use case is SNCF, France’s national state-owned railway service, which operates more than 15,000 trains daily. In its technology suite, the railroad used IBM Watson as part of a long-term plan to apply AI and advanced analytics to railroad operations. The team uses remote sensors to track vibrations, temperature, and pressure. To assist with predictive maintenance, they use sequential machine learning and real-time data processing techniques.