Pixels Attached to Subway Cars for Scientific Research

In a new collaboration with Google, the New York City Metropolitan Transit Authority (MTA) is testing an innovative approach to subway safety. They have attached several Google Pixel phones to subway cars to collect crucial data. This initiative aims to improve the efficiency of track inspections and repairs, which are vital for the daily operation of a subway system that serves over 3 million riders each day.

Traditionally, inspecting subway tracks has relied heavily on human workers. While these inspections are helpful, they often have limitations in terms of speed and thoroughness. Recognizing the need for improvement, the MTA is exploring ways to automate this process and various technologies, including artificial intelligence (AI), that can assist in this area. Reports indicate that Google partnered with the MTA, providing Pixel phones equipped to gather essential data as part of a project called TrackInspect.

The concept behind TrackInspect is straightforward yet groundbreaking. By using commercial smartphones like the Pixel instead of expensive, specialized equipment, the MTA can tap into high-quality data without significant financial investment. The Pixel phones are programmed to listen for track defects and record various other movement data as the subway cars travel along the tracks. They capture sounds like the screeches of the trains and the vibrations of the tracks, which are crucial for identifying potential issues.

This project has highlighted how everyday technology can be repurposed for critical tasks. During the experiment, the Pixel devices collected a tremendous amount of data, including over 335 million sensor readings and 1,200 hours of audio recordings. This information was then used to train around 200 AI models designed to help with the inspection work. The goal is for these AI models to analyze the sounds and vibrations collected by the phones, providing valuable insights that can signal when and where repairs are needed.

While human inspectors are still essential and remain involved in the process, preliminary results are promising. Out of the defects recorded with the Pixel phones, 92% were confirmed by human inspectors, demonstrating a strong correlation between the technology used and actual track issues. The human specialists listened to the audio and examined the vibration data to validate findings with an impressive success rate of 80%.

As the MTA continues to evaluate this technology, there is hope that it can be expanded further. If successful, there may be potential to use more specialized hardware for this purpose in the future, but the initial success of the Pixel phones proves that affordable technology can effectively complement traditional inspection methods. In summary, this initiative represents an exciting step forward for subway safety and track maintenance, utilizing AI and existing technology in a practical and accessible way.