By AI Trends Staff
Driverless trains powered by AI are coming. Driverless train software produced by New York Air Brake was used in a demonstration last summer of a 30-car freight train traveling 48 miles at a research and testing facility owned by the Association of American Railroads, according to a recent account in the Seattle Times.
Today, only in Australia can long, heavy freight trains run without an engineer on board. That effort began when mining company Rio Tinto began running driverless trains to move iron ore hundreds of miles on its remote rail system. A partnership of Australian, Japanese and US companies were involved in developing the software, called AutoHaul, investing more than $940 million 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 are using AI machine learning in many ways outside driverless trains as well.
Texas railroad company BNSF is using AI to improve how it maintains its track and freight cars, according to Muru Murugappan, BNSF Railway chief information officer and vice president of technology services. The company uses machine vision to analyze 750,000 images each day to detect broken wheels, for instance. That system has been deployed in Spokane and is scheduled to be used in North Seattle this year.
“Artificial intelligence has tremendous potential to significantly improve all aspects of our railroad,” Murugappan stated.
The foundation of remote train operations is being put in place with the implementation of Positive Train Control, a system designed to prevent train collisions and speed-related derailments, suggested David Clarke, a professor at the University of Tennessee and director of the school’s Center for Transportation Research. Positive Train Control is required by federal law to be put in place by Dec. 31.
The railroad industry is taking a cue from the development of driverless trucks in the trucking industry. “If the trucking industry drives its costs down by adopting autonomous technology, railroads might feel compelled to do the same,” Clarke stated.
Railroad unions are pushing to have two crew members on board a train at all times. Federal rules do not require it; the debate is ongoing.
Other major technology companies are developing intelligent railroad systems as well. Every mile per hour of speed increase is said to yield $200,000 in benefits, according to a presentation on AI Applications in Rail Transit made at the APTA Rail conference in 2019 from Yousef Kimiagar, former VP, Transit Rail Systems for the Gannett Fleming engineering firm now VP of Advanced Technologies, US Rail Systems for Hatch.)
Benefits of AI in the railroad industry include optimizing and synchronizing timetables to reduce total travel time, waiting time and transfer waiting time, results that increase customer satisfaction and achieve cost savings.
In railroad-targeted software and service options from major technology firms, he cited IBM Smarter Rail, with its dynamic scheduling, surveillance of track and infrastructure and predictive maintenance offerings. Invensys Rail Dimetronic, a train automation supplier serving Spain, is cited in an IBM case study as having helped in the engineering of traffic signaling systems.
Also, GE Movement Planner predicts patterns in train traffic and aims to increase railroad velocity, capacity and efficiency. GE Movement Planning was acquired by Wabtec Corp. in a transaction completed in February 2019. Revenue from GE Movement Planner products was estimated to approach $1 billion in 2019, according to a press release from Wabtec. The transaction was valued at $11.1 billion. Raymond Betler remains president and CEO of the merged company; Rafael Santana, president and CEO of GE Transportation, became president and CEO of Wabtec’s Freight Segment.
A related use case is SNCF, the national state-owned railway company of France, with more than 15,000 daily train runs. The railroad has been using IBM Watson in its technology suite as part of a long-term effort to apply AI and advanced analytics to railroad operations. The team is using remote sensors to detect vibrations, temperature and pressure; it uses sequential machine learning and real-time data processing to help with predictive maintenance.