Between September and In January, six Google Pixel smartphones hitchhiked on four subway cars in New York City. Specifically, they took the train because it lingers 32 miles between the northern end of Manhattan and the southern rivers of Queens.
These phones are not stowaways or vagabonds, and a very sharp passenger can tell because they are inside a plastic shell and are secured to the bottom and interior of the car by brackets. When people inside the car use their smartphones to write emails or scroll Instagram or explore Roblox, subway operators are using sensors from these phones (accelerometers, magnetometers, and gyroscopes), as well as sensors connected to the outside of the car, as well as other external micro machines, other external micro mirrors to listen.
The phones are part of a brief experiment in the Metropolitan Transportation Bureau of New York City, and Google deals with whether cheap, mostly ready-made technology could complement the agency’s track inspection work. (The Google Public Department is the department that does the work and does not charge the MTA for this initial experiment.) Today, human inspectors conducted the inspection, and together they walked 665 miles of all the 665 miles of New York subway tracks, eyes stripped from damaged rails, destructive signals and water damage. Three rides ridden by a dedicated, sensor-based “train geometry” also captures and uploads more complex data on the state of urban rail infrastructure.
Work at the New York City Transportation Company, which Google calls TrackInspect’s experimental technology, shows that audio, vibration and position data are collected relatively cheaply and used to train AI prediction models that can supplement the work of inspection work. It can point humans to a suspicious rattle, bangs or screams, which indicates which tools they need to do for repairs before they get there. The MTA said that throughout the four-month program, the technology was able to determine 92% of the defect locations later precisely pointed out by human track and field inspectors.
Ultimately, the technology could become “we can minimize the amount of work we can identify these defects and point inspectors in the right direction so they can take the time to repair instead of identifying and then go straight there and do the work,” said Demetrius Crichlow, president of the agency. In the future, MTA hopes to create a “modern” system that automatically identifies and organizes fixes for track problems.
For the system’s 3.7 million riders, catching the flaw before a problem occurs can be the difference between going to work or school on time and falling into an unexpected delay.
“This goal [project] Find the problem before it becomes a major issue in terms of service. Crichlow said. The partnership with Google will now expand to a complete pilot project where Google will build a production version of the technology and take it into the hands of track and field inspectors, MTA said.
Inspector’s gadget
Brian Poston, assistant vice president of Transit and Rail, said Google experiments are part of a series of AI-filled technologies that transportation agencies have just begun to use to supplement their typical inspections. New York is unique in using “harmonics” (Audio and vibration), while others have installed small sensors or cameras on tracks that are automatically measured and flag differences when they appear. The technology is not only achieved through advances in machine learning, but also can be cheaper, smaller batteries and processors.
Still, U.S. regulators still need regular inspections and maintenance of rails, and Poston said he does not want the rules to disappear anytime soon. “Until the technology can be specific and precise, you always need this human interaction.”
