Use of Probe Data for Arterial Roadway Travel Time Estimation and Freeway Medium-Term Travel Time Prediction

2015Final report, 7/1/2013 - 12/31/2014PDFTech Reporthttp://www.mautc.psu.edu/docs/MAUTC-2013-05.pdfMAUTC-2013-05DRTR12-G-UTC03AlgorithmsArterial highwaysData qualityEstimatingFloating car dataFreewaysLearning (Artificial intelligence)Mathematical predictionPattern recognition systemsProbe vehiclesTraffic mitigationTravel timeArterial travel time estimationfreeway travel time predictiondata fusionprobe dataparticle filteragent based modelingdynamic template matchingUnited StatesMid-Atlantic Universities Transportation CenterRakha, Hesham A.Chen, HaoHaghani, AliZhang, XuechiHamedi, MasoudVirginia Polytechnic Institute and State UniversityVirginia Tech Transportation InstituteUniversity of Maryland (College Park, Md.). Dept. of Civil and Environmental EngineeringUnited States. Dept. of Transportation. Research and Innovative Technology AdministrationMid-Atlantic Universities Transportation CenterUS Transportation CollectionUrban traffic congestion is a problem that plagues many cities in the United States. One approach to alleviating this congestion is to provide drivers with better travel time information so that they can make better departure time and routing decisions. This research project focuses on two efforts: (1) validating the use of probe data to estimate arterial travel times and (2) validating and developing techniques to predict freeway travel times over a 2 to 4 hour window. With regards to the arterial travel time estimation the objective are two-fold. The first goal is a comprehensive validation of INRIX arterial data. In the validation the variability and reliability of arterial data in different corridors equipped with permanent and portable traffic sensors will be studied under different traffic conditions. The second goal is to develop a methodology for augmenting INRIX data with other data sources in order to improve the data quality. With regards to freeway travel time prediction the objectives are also two-fold. The first objective is to validate the INRIX travel time prediction algorithms along a number of corridors in the state of Virginia. The second objective is to enhance the prediction accuracy using various pattern recognition and machine learning techniques