Exploratory Methods for Truck Re-identification in a Statewide Network Based on Axle Weight and Axle Spacing Data to Enhance Freight Metrics

Importance of freight transportation is recognized universally by all states and it is particularly an important component of Oregon’s economy. Many sources estimate that freight shipments will nearly double in 10 years. While other modes are clearly important for freight transportation, trucking is the dominant mode in terms of tons and value. Monitoring freight movement and freight transportation performance is essential in making effective policies and informed decisions to enhance and efficiently manage the freight transportation system. One of the key aspects of monitoring freight over the highways has to do with determining the flow patterns of trucks, which can…

Importance of freight transportation is recognized universally by all states and it is particularly an important component of Oregon’s economy. Many sources estimate that freight shipments will nearly double in 10 years. While other modes are clearly important for freight transportation, trucking is the dominant mode in terms of tons and value. Monitoring freight movement and freight transportation performance is essential in making effective policies and informed decisions to enhance and efficiently manage the freight transportation system.

One of the key aspects of monitoring freight over the highways has to do with determining the flow patterns of trucks, which can be achieved by uniquely identifying trucks at specific points along the roads or by tracking individual trucks using technology such as GPS. For example, in Oregon, there are 20 active reporting and equipped stations where trucks carrying Green Light transponders can be uniquely identified. These stations also include WIM (weigh-in-motion) systems which provide axle weights, spacing, and gross vehicle weight estimates uniquely matched to a transponder-equipped truck. This proposed research seeks to develop an a new method to determine flow patterns of trucks by matching archived vehicle-attribute data such as axle spacing and axle weights at multiple geographic locations. The proposed method, in some circumstances, can be more advantageous as compared to other available options to track and re-identify trucks (e.g., GPS, automatic vehicle identification (AVI), license plate recognition) because of several reasons. i) Data from AVI transponders, such as Green Light, or from other types of electronic tracking systems might not be readily available to the public agencies involved in motor freight planning (e.g., MPOs, DOTs) due to privacy, jurisdictional, and institutional issues. ii) Not all trucks are equipped with AVI transponders. However, with the proposed methods all trucks can be potentially tracked since they all cross the WIM stations. iii) The proposed approach does not require installation of any new sensors since the input data are already collected at existing WIM and automatic vehicle classification (AVC) stations, whereas alternative technologies like license plate recognition requires additional investment.

Overall, this research focuses on 1) developing advanced methods and algorithms to anonymously identify and match commercial trucks crossing two data collection stations on roadways; and 2) on investigating how these re-identification methods can be employed to enhance freight metrics. The methods for truck re-identification will utilize only vehicle-attribute data such as axle spacing and axle weights that are typically collected by AVC and WIM sensors installed on many roadways. By capitalizing on the vehicle-attribute data from a number of AVC and/or WIM stations in a network, the proposed methods can potentially support and benefit multiple applications, such as determining travel times, quantifying travel time reliability, estimating truck flow patterns (i.e., origins-destinations), estimating empty truck movements, trip length estimation, tracking movements of trucks without transponders, and pavement management.

This research builds on existing exploratory work in a current OTREC project and previous work by co-PIs using WIM data from Indiana. The PIs have already compiled extensive data from 20 WIM stations in Oregon that are currently recording data. These data will be used for model development, validation, and testing. The co-PIs have completed preliminary work on truck re-identification based on axle spacing and weight data from Indiana, and showed that it is feasible to successfully match trucks crossing two WIM stations that are separated by one mile. In Oregon, there are 22 WIM stations across the state. This will allow the research team to investigate and develop sophisticated methods to match trucks crossing WIM stations separated by larger distances in a network setting. New methods will be developed to account for varying truck travel times between two stations (e.g., due to stops for fuel, rest, or deliveries). In addition, the performance of the matching algorithms will be tested when only axle spacing data are used as input to evaluate the matching accuracy if only AVC sensors are available at the stations. To develop the necessary methods for this study, state-of-the-art mathematical optimization and statistical modeling techniques, such as Bayesian methods, finite mixture models, nonlinear optimization, and neural networks, will be investigated and employed. Extensive analyses will be performed to clearly understand how such systems would perform in real-world applications. The results of this study will benefit not only Oregon but potentially all other states since truck characteristics do not vary significantly from state to state, and many states also collect axle spacing and axle weight data.