Title: Crowdsourcing Bridge Vital Signs with Smartphone Vehicle Trips

Abstract: The efficacy of sensor data in modern bridge condition evaluations has been undermined by inaccessible technologies. While the links between vibrational properties and structural health have been well established, high costs associated with specialized sensor networks have prevented the integration of such data with bridge management systems. In the last decade, researchers predicted that crowd-sourced mobile sensor data, collected ubiquitously and cheaply, will revolutionize our ability to maintain existing infrastructure; yet no such applications have successfully overcome the challenge of extracting useful information in the field with sufficient precision. Here we fill this knowledge gap by showing that critical physical properties of a real bridge can be determined accurately from everyday vehicle trip data. We collected smartphone data from controlled field experiments and UBER rides on the Golden Gate Bridge and developed an analytical method to recover modal properties, which paves the way for scalable, cost-effective structural health monitoring based on this abundant data class. Our results are consistent with a comprehensive study on the Golden Gate Bridge. We assess the benefit of continuous monitoring with reliability models and show that the inclusion of crowd-sourced data in a bridge maintenance plan can add over fourteen years of service (30% increase) to a bridge without additional costs. These results certify the immediate value of large-scale data sources for studying the health of existing infrastructure, whether the data are crowdsensed or generated by organized vehicle fleets such as ridesourcing companies or municipalities.