What you need to know
- Google and the New York Metro Transit Authority came together to perform an experiment to gain insights on track safety.
- They installed several Google Pixel phones inside and under subway cars, which could help them track movement and listen for any track defects.
- This experiment, dubbed ‘TrackInspect,’ significantly lowers the need for human-led inspections as the Pixels would capture audio, vibration, and location data.
Imagine a world where there were no subway service interruptions due to scheduled or unexpected repairs on the rail tracks. This is exactly what the Metropolitan Transportation Authority and Google are hoping to achieve in the future with their pilot program.
Google and the New York Subway took part in an experiment that involved sticking pixel phones on the inside of subway cars as well as underneath them (via Wired).
This experiment was conducted between September 2024 and January 2025 on four New York City subway cars by Google’s Public Sector division, with trains shuttled between Manhattan and the south of Queens. The aim was to see if phones could possibly replace or lower the need for human-led inspection along roughly 665 miles of New York City’s subway.
The publication added that the subway operators were equipped with “accelerometers, magnetometers, and gyroscopes” along with the Google Pixels. At the same time, the ones attached on the outside had additional microphones to go with them. Usually, all this keen listening and observing is done manually by inspectors.
Google seemed to be working on a special tech that would equip standard Pixel phones to perform more complex tasks like the above. Dubbed TrackInspect, this is set up to capture sound and vibration data, which is sent to a cloud-based system. These systems have AI capabilities that will then use this data to predict subway track defects.
Breaking it down, the MTA (Metropolitan Transportation Authority) states that the prototype “sends a soundbite or noise clip showing heavy vibration or noise, and then our inspectors follow up by walking the track and verifying any issue found.”
The results of this experiment seemed to be extremely insightful. According to Brent Mitchell, Google’s Public Sector vice president, 92% of the defects spotted by the TrackInspect program were similar to those tracked by human-led inspections. MTA stated that throughout this experiment, TrackInspect collected 335 million sensor readings, one million GPS locations, and 1,200 hours of audio, which was later ingested into a machine-learning model running on Google Cloud.
With this initiative, the MTA wants to look at possible ways of conducting safer inspections of miles-long subway tracks in the future, also fixing track issues faster without much disruption to their services. It is always great when technology carefully monitored by humans works its magic, making things more cost-effective and efficient while paving the way for other industries to consider this route as well.
While we don’t know which generation of Pixels were used in this experiment, Android Central has reached out to Google for more insights about the equipment used and will update this article once we have more information.