RESEARCH INITIATION AWARD GRANT: STATISTICAL MODELS FOR REAL-TIME PARAMETER PREDICTION FROM PROBE VEHICLE DATA AT TRAFFIC INTERSECTIONS

Benedict College, Columbia SC

Investigators

Abstract

The objectives of the project entitled: Statistical Models for Real-Time Parameter Prediction from Probe Vehicle Data at Traffic Intersections, are to develop new queue length estimate models at traffic intersections, to teach undergraduate students mathematical and computational modeling skills, and to develop project based undergraduate courses. The project will investigate the estimation of system parameters such as queue length, delay, and flow rate at traffic intersections based on data from vehicles instrumented with wireless communication and location tracking technologies, so called probe vehicles. Estimating system states in real-time can enable optimal control through efficiently allocating the available capacity such that a defined performance metric is optimized, for example the queue length is minimized. To estimate these performance measures in real time, various surveillance technologies are being employed to measure traffic flow parameters which are subsequently utilized in models for delay estimation and prediction. However, these detection technologies are not effective in estimating queue lengths or delays. Research is needed to develop models using probe information such as count, location, and time, and to understand how probe vehicle technology could potentially improve the estimation of desired parameters. During this project,first queue length estimators for two-lane isolated intersections with left and right turn movements for Poisson arrivals, then platoon arrivals will be developed. Fixed and random service time distributions will be investigated. Finally, developed models will be tested with VISSIM microscopic simulation platform in order to show if the models are accurately estimating the desired parameters and under what conditions the models are adequate.