Crowdsourcing Bridge Vital Signs with Smartphone Vehicle Trips

Modern bridge health monitoring methods require specialized sensor networks, which have costs that are prohibitive to bridge owners. Mobile sensor networks reduce costs by capturing vibrational signatures and indicators of structural decay using substantially fewer devices. Over the last decade, researchers have hypothesized that crowd-sourced mobile sensor data, collected ubiquitously and cheaply, will revolutionize our ability to maintain existing infrastructure. In this context, crowd-sourced data faces a precision challenge in which magnitude-level leaps must be made over past applications. As such, the field is missing a realistic field validation of this concept; researchers have primarily focused on synthetic models, controlled experiments, and idealized measurements. Here we fill this knowledge gap by showing that critical physical properties of a real bridge can be determined 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 dominant vibrational frequencies, 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; they 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