Riders of the New York City subway are all too familiar with its numerous issues, ranging from unusual noises to combustible debris on the tracks. In an effort to address these problems, the Metropolitan Transportation Authority (MTA) is currently exploring the potential of AI-driven solutions to enhance the repair process, in collaboration with six Google Pixel phones.

Between September and January, six Google Pixel phones were installed on four different subway cars to collect data. This experiment, conducted in partnership with Google Public Sector, utilized the phones’ accelerometers, magnetometers, and microphones to detect suspicious noises. The collected data was then transmitted to cloud-based systems, which applied machine learning algorithms to generate predictive insights.

The technology, known as TrackInspect, successfully identified 92 percent of the defect locations that human inspectors had found. According to New York City Transit President Demetrius Crichlow, “By detecting early defects in the rails, we can save both money and time for our crew members and riders.” In a press release, he emphasized that this innovative program, the first of its kind, leverages AI technology to not only provide a smoother ride for customers but also enhance the safety of track inspectors by equipping them with more advanced tools.

Traditionally, inspectors manually walk the entire 665 miles of subway tracks to identify issues, supplemented by sensor-equipped “train geometry cars” that collect data three times a year. During the experiment, inspectors verified the locations highlighted by the AI system and confirmed the presence of defects. They could also inquire about maintenance protocols and ask questions through the system’s generative AI interface.