The Subway System in New York City Utilizes Google Pixels to Detect Track Defects

Between September and January, six Google Pixel smartphones took unexpected trips on four subway cars in New York City, specifically riding the A train, which travels 32 miles from the north of Manhattan to the south of Queens. These phones were not simply lost or abandoned; they were securely enclosed in plastic and attached to the cars’ undersides and interiors. While subway passengers were busy on their own devices—checking emails, scrolling through Instagram, or playing games—the subway operators were putting these phones’ sensors to good use. The phones contained accelerometers, magnetometers, and gyroscopes, and on the outside cars, they had additional microphones to pick up sounds.

The smartphones were part of an experiment conducted by the Metropolitan Transportation Authority (MTA) of New York City in collaboration with Google. The goal was to see if this inexpensive, mainly off-the-shelf technology could help with the MTA’s track inspection processes. Traditionally, human inspectors walk all 665 miles of subway tracks, looking for problems like broken rails, faulty signals, or water damage. Specialized “train geometry cars” also make rides three times a year to collect more detailed data about the rail system’s condition.

The initial project, called TrackInspect by Google, showed promising results. By collecting audio, vibration, and location data, the technology could assist human inspectors in identifying issues before they grew into bigger problems. This approach could help pinpoint strange noises or vibrations, indicating what kind of tools inspectors would need before they arrived at the site. According to the MTA, during the four-month testing period, the system successfully identified 92 percent of the problem locations that human inspectors later confirmed.

Demetrius Crichlow, the president of the agency, mentioned that this technology could help reduce the time spent on identifying rail defects. Instead of searching for issues, inspectors could focus on making repairs. The MTA envisions a future where a modernized system automatically spots and prioritizes track issues for quick resolution. For the millions of daily subway riders—currently around 3.7 million—catching these defects early could mean the difference between arriving on time and facing unexpected delays.

The primary aim of this project is to tackle problems before they escalate into significant service interruptions. Building on the success of the initial experiment, the MTA plans to extend the partnership with Google into a full pilot project. Google will develop a more refined version of this technology that will be used directly by track inspectors.

This initiative is part of a growing trend where transit agencies are beginning to incorporate AI-powered tools to enhance their usual inspection processes. Although New York’s approach—using audio and vibrations to identify issues—is somewhat unique, other agencies are also using small sensors or cameras on tracks to automate measurements and identify problems as they arise. Advances in machine learning, along with the availability of smaller and more affordable batteries and processors, have made these technologies more accessible.

However, U.S. regulators still mandate regular human inspections and maintenance for rail systems, and experts believe this requirement will remain in place for the foreseeable future. According to Brian Poston, an assistant vice president at WSP, full reliance on technology will take time, as precise and specific measurements are still necessary. Until such advances are achieved, human oversight will remain essential in maintaining the safety and efficiency of public transportation.