Characterizing Pavement Roughness at Non-Uniform Speeds Using Connected Vehicles

Abstract

Methods of pavement roughness characterizations using connected vehicles are poised to scale beyond the frequency, span, and affordability of existing methods that require specially instrumented vehicles and skilled technicians. However, speed variability and differences in suspension behavior require segmentation of the connected vehicle data to achieve some level of desired precision and accuracy with relatively few measurements. This study evaluates the reliability of a Road Impact Factor (RIF) transform under stop-and-go conditions. A RIF-transform converts inertial signals from on-board accelerometers and speed sensors to roughness indices (RIF-indices), in real-time. The case studies collected data from 18 different buses during their normal operation in a small urban city. Within 30 measurements, the RIF-indices distributed normally with an average margin-of-error below 6%. This result indicates that a large number of measurements will provide a reliable estimate of the average roughness experienced. Statistical t-tests distinguished the relatively small differences in average roughness levels among the roadway segments evaluated. In conclusion, when averaging roughness measurements from the same type of vehicle moving at non-uniform speeds, the RIF-transform will provide everincreasing precision and accuracy as the traversal volume increases.