Proactive Safety Improvement via crowdsourcing Live Curve Safety Assessment

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

This project will create novel methodologies for assessing, managing, and improving the safety conditions of horizontal curves on roadway networks. In the US, a disproportionately high number of serious vehicle crashes occur on horizontal curves (25 percent of all fatal crashes), despite curves accounting for only 5 percent of total highway mileage. The advisory speed of curves can change due to pavement condition changes caused by pavement aging and distress, and curve geometry change can be caused by resurfacing and maintenance. This can lead to a higher risk of crashes if the curve advisory speed has not been assessed, monitored, and adjusted in a timely manner. However, the current network-level curve safety assessment practices are labor-intensive, time-consuming, and costly for engineers, resulting in problematic curves often being identified only after crashes occur. To address this issue, this project aims to develop new methodologies that leverage low-cost mobile devices, intra-agency crowdsourcing, advanced and scalable algorithms, and a novel confidence-driven, decision-making method that will make cost-effective and frequent curve safety assessments technically and economically feasible. The methodologies that will be developed in this project will result in significant savings in terms of cost and labor for transportation agencies and will promote transportation safety equity, especially for agencies with limited resources. The research outcomes will lay a solid foundation for advancing the development of future technologies and tools, and, most importantly, they will take a big step towards development of a safe system that saves lives by transforming the reactive safety management practices into proactive roadway safety management practices. The objectives of this project are to develop and validate two methodologies for network-level curve safety assessment by taking a multidisciplinary approach. The first methodology is a crowdsourced multi-run curve safety assessment methodology that will address the technical challenges in the state-of-art technologies related to different vehicle types, suspension properties and driver behaviors when scaling up from individual curve safety assessment to network-wide assessment with a crowdsourcing framework. This methodology will use a novel seed-propagation model and an iterative convergence feedback model combined with the curve-driving kinematics relationship and spatial-temporal analysis for simultaneous estimation of vehicle suspension and curve geometric properties. The second methodology is a trajectory and roadway geometry-dependent confidence level methodology that will enable informed, data-driven decision-making by quantitatively evaluating the confidence level of the computed decision-making outcomes, such as advisory speed, based on a vehicle’s trajectory and roadway geometry. The outcomes of this project will also be integrated into a wide range of classroom and research activities to train the next generation of scientists, engineers, and policymakers on innovative technologies that will promote proactive safety management practices and transportation safety equity. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.