Precision Bounds of Pavement Distress Localization with Connected Vehicle Sensors

Abstract

Continuous, network-wide monitoring of pavement performance will significantly reduce risks and provide an adequate volume of timely data to enable accurate maintenance forecasting. Unfortunately, transportation agencies can afford to monitor less than 4% of the nation’s roads. Even so, agencies monitor their ride quality at most once annually because current methods are expensive and laborious. Distributed mobile sensing with connected vehicles and smartphones could provide a viable solution at much lower cost. However, such approaches lack models that improve with continuous, high-volume data flows. This research characterizes the precision bounds of the road impact factor transform that aggregates voluminous data feeds from geoposition and inertial sensors in vehicles to locate potential road distress symptoms. Six case studies of known bump traversals reveal that vehicle suspension transient motion and sensor latencies are the dominant factors in position estimate errors and uncertainty levels. However, for a typical vehicle mix, the precision improves substantially as the number of traversals approaches 50.