Abstract
Understanding truck activity is an essential component of strategic freight planning and analysis. Current truck activity data sources available in the United States include telematics from technologies such as global positioning systems (GPS), axle-based classification counts traffic detector infrastructure and survey data from mail and intercept surveys. GPS data can be used to provide detailed truck trajectories for a subset of the truck population; axle-based classification counts provide location specific volumes by according to axle-based configuration; and survey studies can reveal travel patterns of specific truck vocations from limited samples. Although truck body configurations are a good indicator of industry affiliation and in some cases types of commodities hauled, existing activity data are not available at such a level of detail. Hence, none of these data sources can be used to provide freight truck activity across key corridors, cordons and gateways. As a consequence, a significant data gap still exists in obtaining freight- and commodity- specific truck activity on the road network to understand their temporal and spatial patterns. The University of California, Irvine, Institute of Transportation Studies (UCI-ITS) recently developed a new state-wide data source called the Truck Activity Monitoring System (TAMS, http://freight.its.uci.edu/tams) by enhancing existing inductive loop-based traffic detector infrastructure across the State of California with inductive loop signature technology. Data for the model development effort was collected at over a dozen locations in California to capture a full representation of truck body configurations. Detailed body configuration classification models were subsequently developed from this extensive dataset using a combination advanced machine learning algorithms of naïve Bayes, decision trees, support vector machines and artificial neural networks, integrated through an ensemble learning framework. The outcome is a model that is capable of predicting over 40 different truck body configurations using inductive signature data generated from only a single loop sensor – the most ubiquitous traffic sensor in the State of California. Because of the inductive loop sensor’s extensive state-wide implementation, the classification system developed from this study was deployed along major truck corridors by updating existing loop detector sites at over 90 locations across the State of California, encompassing state gateways, inter-regional cordons and key metropolitan corridors. Data from these sites are transmitted wirelessly to a central database in UCI-ITS and processed automatically to yield hourly summaries of detailed truck volumes. An interactive GIS-based website was developed to provide users with advanced visual analytics and access to archived data across all deployed locations. The data generated by this system is being used for the validation of the California State-wide Freight Forecasting Model, the Caltrans Traffic Census Program, and is being used for regional truck activity models such as the SCAG Heavy Duty Truck Model.
Speaker Bio
Dr. Andre Tok is Testbeds Manager and Assistant Project Scientist at the University of California, Irvine, Institute of Transportation Studies (UCI-ITS). He obtained his Ph.D. in Civil and Environmental Engineering from UCI-ITS in 2008. Dr Tok’s is passionately interested in addressing data gaps in freight transportation analysis and modelling. His research accomplishments include the development of the Online California Freight Data Repository, the California State-wide Freight Forecasting Model, and the California Vehicle Inventory and Use Pilot Survey. His research team recently completed the California Truck Count Study where they developed a state-wide truck classification system based on inductive signature technology to improve truck activity data for freight analysis. His current research involvement includes leading a survey study to identify medium-heavy duty truck vocations that will benefit from aerodynamic retrofits, investigation of integrating wireless technologies with inductive signature systems to improve truck activity data, and investigating emissions benefits of replacing conventional diesel with natural gas trucks.
For more information about the ESD Seminars Series, please contact Karthik Natarajan at karthik_natarajan@sutd.edu.sg.
