| dc.description.abstract | Advances in computing capabilities, image processing, and sensing technologies have permitted the development of specialized vehicles equipped with the capability to assess pavement condition at normal operating speeds. This has greatly improved engineers’ ability to assess and manage pavements, but the equipment is costly, and not all agencies can afford to purchase it. Recently, researchers have developed smartphone applications to address this data collection problem, but most of this work focused on a restricted setup, or calibration. This paper presented a methodology to estimate a ride quality index (RQI) from crowdsourced smartphone measurements and validated this approach with the use of statistical methods. This investigation was divided into three phases. First, a mechanical model to assess ride quality was developed. Second, the Monte Carlo method and the probabilistic point estimate were adopted to simulate RQI measurement responses to different longitudinal profiles from different vehicle traffic spectra. Third, the effects of wander and multilane effects in estimating the minimum required sample size for RQI measurements to converge were evaluated. Once the mechanical model was developed, the results from the Monte Carlos simulations showed that in 83% of cases, the RQI measurements showed no statistical significance. The results from the effect of multilane and wandering effects showed that the sample size for RQI measurements to converge adopting a coefficient of variation of 2% is 400 samples considering a single lane and wander, and 435 samples considering two lanes and wander. The use of the Monte Carlo method successfully validated the crowdsourced smartphone-based RQI measurements as an alternative method to evaluate pavement condition. This approach has the potential to save transportation agencies millions of dollars in pavement condition surveys and to give a better sense of pavement condition in real time. | |
