Crowdsourcing Bridge Vital Signs with Smartphone Vehicle Trips

2022: Announcement

On November 6, 2022, the Massachusetts Institute of Technology announced the development of an application with which smartphones installed in cars can collect data on the integrity of structures when crossing bridges. Thus, they can become a less expensive alternative to sets of sensors installed on the bridges themselves.

The main conclusion is that information on the state of bridge structures can be extracted from accelerometer data collected by a smartphone, says Carlo Ratti, director of MIT Sensable City Laboratory and co-author of the study.

Application that determines the risks of bridge collapse from smartphone vibrations

To conduct the study, scientists developed the Crowdsourcing Bridge Vital Signs with Smartphone Vehicle Trips mobile application to collect accelerometer data when the devices were installed on cars passing over the bridge. They could then look at how well that data matched data recorded by sensors on the bridges themselves to see if the mobile phone method worked.

In our work, we developed a technique for extracting vibration modal frequencies from noisy data collected from smartphones, "says study author Paolo Santi, chief scientist at MIT's Senseable City Lab. As data from multiple bridge trips is recorded, the noise generated by the engine, suspension vibration, traffic, [and] asphalt tends to be leveled off, while the main dominant frequencies appear.

In the case of the Golden Gate Bridge, researchers drove 102 times over the bridge with devices on, and also used 72 rides in Uber cars with phones on. The team then compared the findings with data from a group of 240 sensors that had been placed on the Golden Gate Bridge for three months.

As a result, it turned out that data from smartphones converge with data from bridge sensors; for the 10 specific types of low-frequency vibrations that engineers measure on the bridge, a close match was observed, and in five cases there was no discrepancy between the methods at all.

Our first results suggest that only [a modest number] of journeys over a few weeks are sufficient to provide useful information on modal bridge travel frequencies, says Santi.[1]

Notes