New York launches AI-driven track maintenance pilot

The Metropolitan Transportation Authority (MTA) has launched a pilot programme in partnership with Google Public Sector to improve subway track maintenance through the use of artificial intelligence (AI) and sensor technology.

The TrackInspect initiative integrates sensor hardware with cloud-based systems to detect potential track defects before they result in service disruptions.

“The TrackInspect complements the current practices by supplementing the existing inspections with valuable targeted data based on trains traversing areas and access to resources at their fingertips while in the field utilising the conversation interface,” an MTA spokesperson told Cities Today.

The programme uses Google Pixel smartphones retrofitted with standard plastic cases and mounted on R46 subway cars along New York’s A line. These smartphones capture vibrations and sound patterns through built-in sensors and microphones, identifying areas that may require preventive maintenance.

The data collected is transmitted in real-time to cloud-based systems where machine learning algorithms analyse the information and generate predictive maintenance alerts. Track inspectors then follow up on flagged locations to confirm the condition and ensure maintenance is performed.

“Currently, there are track inspectors that walk the track twice a week, visually looking for track defects,” added the spokesperson. “There is also a Track Geometry Car that can identify defects and supervisors go out to the field to verify and repair those defects before they become a condition that can impact operations.”

MTA reports that the programme has already shown positive results, identifying 92 percent of defect locations found by track inspectors during the pilot phase. During the first phase, the system collected 335 million sensor readings, 1 million GPS locations, and 1,200 hours of audio data.

This data, combined with information from the MTA’s track geometry cars, enhances the speed and accuracy of track diagnostics, helping to identify and address track issues more efficiently.

In response to questions about the scalability and cost of expanding the programme, an MTA spokesperson explained: “The technology has just been launched as a pilot programme to better understand if and how this technology could be further refined and implemented at a larger scale.”

The MTA is also exploring other sensor and analytics solutions that could integrate into the TrackInspect system, with a focus on ensuring interoperability between various technologies.

Image: Kit1nyc | Dreamstime.com