This poster session contains papers that describe advances and innovations in monitoring passenger vehicle traffic, truck traffic, and pedestrian traffic. It includes two papers focused on monitoring traffic during the COVID-19 pandemic.
Using High-Resolution Archived Operational Traffic Data for Transportation Management during COVID-19 Pandemic
Yang Cheng (firstname.lastname@example.org), University of Wisconsin, MadisonShow Abstract
Keshu Wu, University of Wisconsin, Madison
Steven Parker, University of Wisconsin, Madison
Elizabeth Schneider, Wisconsin Department of Transportation
Jonathan Riehl, University of Wisconsin, Madison
David Noyce, University of Wisconsin, Madison
The COVID-19 pandemic has led many governmental agencies to issue safer-at-home orders, and many employers have also implemented work from home strategies, as efforts to contain the spread of the disease. Those measures initially led to significant reductions in travel demand and traffic daily volumes. Most traffic management strategies, however, are designed and calibrated based on historical prevailing traffic conditions. Therefore, traffic management agencies are looking for available data sources to better understand the rapid and unprecedented changes and modify their control strategies if needed. This study presents use cases for using high-resolution traffic data from state fixed detectors for traffic control and law enforcement practices: (1) ramp metering schedule adjustment, and (2) speeding hotspot identification. This study further investigates the speeding behaviors and finds that although speeding does exist at certain locations, most travelers tended to drive at a lower speed during the safer-at-home order, and the speed deviations among traffic were less than before the pandemic. This study also demonstrates that high-resolution fixed location detector data (1-minute or less) is able to provide more insights while traditionally used data (aggregated in five minutes or longer) could not.
Examining Impacts of COVID-19 Related Stay-at-Home Orders through a Two-way Random Effects Model
Anshu Bamney, Michigan State UniversityShow Abstract
Hisham Jashami, Michigan State University
Sarvani Sonduru Pantangi, Michigan State University
Jayson Ambabo, Michigan State University
Megat-Usamah Megat-Johari, Michigan State University
Qiuqi Cai, Michigan State University
Nischal Gupta, Michigan State University
Peter Savolainen, Michigan State University
The COVID-19 pandemic has had far-reaching impacts on public health and safety, economics, and the transportation system. In response to this disease, social distancing has been identified as an important means to reduce the potential for transmission of the disease. To this end, federal and local governments around the world have introduced stay-at-home orders and other restrictions on travel to ‘non-essential’ businesses. Preliminary evidence suggests substantial variability in the impacts of these orders in the United States, both across states and over time. This study examines this issue using daily county-level vehicle-miles traveled (VMT) data for the 48 continental U.S. states and the District of Columbia. A two-way random effects model is estimated to assess changes in VMT from March 1 to June 30, 2020 as compared to baseline January travel levels. The introduction of stay-at-home orders was shown to significantly reduce VMT, particularly during the early stages of the pandemic. However, over time, these effects were shown to dissipate, which may be attributable to ‘quarantine fatigue’. Many of the stay-at-home orders were lifted, in whole or in part, and travel was also shown to be lower when restrictions were in place on certain non-essential businesses, particularly those related to retail and personal care. Travel was also lower in those areas where higher numbers of COVID-19 cases were reported, regardless of the status of such restrictions. Traffic levels also varied with respect to other characteristics, including median income, political leanings, and how rural the county was in nature.
Data-Driven Approach to Quantify and Reduce Error Associated with Assigning Short Duration Counts to Traffic Pattern Groups
Giuseppe Grande, University of ManitobaShow Abstract
Puteri Paramita, University of Manitoba
Jonathan Regehr, University of Manitoba
Traffic monitoring agencies collect traffic data samples to estimate annual average daily traffic (AADT) at short duration count sites. The steps to estimate AADT from sample data introduce error that manifests as uncertainty in the AADT statistic and its applications. Past research suggests that the assignment of a short duration count site to a traffic pattern group (TPG), characterized by known traffic periodicities, represents a significant but poorly quantified source of error. This paper presents an approach to quantify the range of errors arising from such assignments and to mitigate these errors using a novel data-driven assignment method. The approach uses simulated 48-hour short duration counts sampled from continuous count sites with known AADT to develop a benchmark of the total error expected when AADT is estimated from such samples. Likewise, the analysis produces a set of AADT estimates using temporal factors from pre-defined TPGs to quantify the range of assignment errors. The data-driven assignment method aims to mitigate these errors by minimizing the variance in AADT estimates produced from multiple short duration counts in a single year. The approach is applied to traffic data collected in Manitoba, Canada, as a case study. The results indicate that the mean absolute error from 48-hour short duration counts is 6.40% of the true AADT and that improper assignment can lead to a range in mean absolute errors of 9%. When applied to previously unassigned sites, the data-driven assignment method reduced mean absolute errors from 10.32%, using a conventional assignment method, to 7.86%.
Multi-Sensor Data Fusion for Accurate Traffic Speed and Travel Time Reconstruction
Lisa Kessler (email@example.com), Technical University of MunichShow Abstract
Felix Rempe, BMW Group
Klaus Bogenberger, Technische Universitat Munchen
This paper studies the joint reconstruction of traffic speeds and travel times fusing sparse sensor data. Raw speed data from inductive loop detectors and floating cars as well as travel time mea surements are combined using different fusion techniques. A novel fusion approach is developed which extends existing speed reconstruction methods to integrate low-resolution travel time data. Several state-of-the-art methods, the novel methods and combinations of sensor data are evaluated with respect to their performance in reconstructing traffic speeds and travel times. Algorithms and sensor setups are evaluated with real loop detector, floating car and Blue tooth data collected during a severe congestion on German freeway A9. Two main aspects are examined: (i) which algorithm provides the most accurate result depending on the used data and (ii) which type of sensor and which combination of sensors yields higher estimation accuracies. Results show that, overall, the novel approach applied to a combination of floating-car data and loop data provides the best speed and travel time accuracy. Furthermore, a fusion of sources improves the reconstruction quality in many, but not all cases. Only if integrated distinctively, Bluetooth data provide a benefit for reconstruction purposes.
Low Cost Two Dimensional (2–D) LiDAR Application for Vehicle Trajectory Construction at The Intersections
RAVI JAGIRDAR (firstname.lastname@example.org), JPCL Engineering LLCShow Abstract
Joyoung Lee, New Jersey Institute of Technology
Dejan Besenski, New Jersey Institute of Technology
Min-Wook Kang, University of South Alabama
This paper presents a vehicle detection and tracking methodology, using a low-cost two-dimensional (2-D) LiDAR data, to effectively obtain the turning movement counts at intersections. The proposed methodology integrates a clustering technique, an inverse sensor model, and Kalman filter to render the trajectories of individual vehicles at intersections. The objective of applying K-means clustering is to robustly differentiate LiDAR data generated by pedestrians, vehicles, and the presence of multiple vehicles. An inverse sensor model is a state model of occupancy grid mapping that localizes the detected vehicles on the grid map with known LiDAR position. A constant velocity model based Kalman filter is defined to construct the trajectory of the vehicles. The data are collected from two intersections located in Newark, New Jersey, to examine the proposed method's accuracy. The results show that the proposed method has an average accuracy of 83.8%. Furthermore, the obtained R-squared value for localizing the vehicles on the grid map ranges from 0.87 to 0.89. Additionally, the accuracy of the proposed method is compared with current data collection technology. Furthermore, a primary cost comparison is made to study the cost efficiency of the developed methodology. The cost comparison shows that the proposed methodology based on 2-D LiDAR technology can achieve acceptable accuracy at a low price and be considered a smart city concept to conduct extensive scale data collection. Keywords: 2 -D LiDAR, Turning Movement Counts, K-Means Clustering, Inverse Sensor Model, Kalman Filter
On Spatial Transferability of Machine Learning based Volume Estimation Models
Kevin Kasundra, National Renewable Energy Laboratory (NREL)Show Abstract
Venu Garikapati, National Renewable Energy Laboratory (NREL)
Christopher Hoehne, National Renewable Energy Laboratory (NREL)
Yi Hou, National Renewable Energy Laboratory (NREL)
Stanley Young, National Renewable Energy Laboratory (NREL)
High-quality traffic volume data is essential for efficient transportation planning and operations. However, such high-quality data is expensive to collect, owing primarily to the high capital cost of installing and maintaining continuous counting stations (CCSs). Recent availability of probe-based vehicle data offers a cost-effective solution for increasing the observability of traffic volumes. However, having ample ground truth traffic data is a prerequisite for developing robust volume estimation models. Though this might not be a big issue in many states, states with scarce CCS data might be able to benefit from robust volume estimation models developed in (adjacent) data-rich states. While there is a reasonable amount of spatial transferability research in the transportation domain, there is a dearth of knowledge on the spatial transferability of probe-based volume estimation models. To address this gap, this paper explores spatial transferability of volume estimation models developed from data in three states (Colorado, North Carolina, and Pennsylvania). Results indicate that it is extremely important to maintain temporal consistency when attempting spatial transferability of volume estimation models. It was also found that models trained on regions with lower peak traffic volumes will limit the performance of models transferred to states with higher peak hourly traffic volumes. Corroborating findings from existing spatial transferability research on other topics, it was found that a meta-model (developed using data from multiple states) performs better than volume estimation models developed within any one of the states.
An Ensemble Approach to Truck Body Type Classification using Deep Representation Learning on 3D Point Sets
Yiqiao Li (email@example.com), University of California, IrvineShow Abstract
Koti Allu, University of California, Irvine
Zhe Sun, University of California, Irvine
Andre Tok, University of California, Irvine
Guoliang Feng, University of California, Irvine
Stephen Ritchie, University of California, Irvine
Understanding spatiotemporal distribution of commercial vehicles is essential for facilitating strategic pavement design, freight planning, and policy making. Hence, transportation agencies have been increasingly interested in collecting truck body configuration data due to their strong affiliation with industries and freight commodities to better understand their distinct operational characteristics and impacts on infrastructure and the environment. The rapid advancement of the Light Detection and Ranging (LiDAR) technology has facilitated the development of non-intrusive detection solutions that are able to accurately classify truck body types in detail. This paper proposes a new truck classification method using a LiDAR sensor oriented to provide a wide field-of-view of roadways. In order to enrich the sparse point cloud obtained from the sensor, point clouds originating from the same truck across consecutive frames were grouped and combined using a two-stage vehicle reconstruction framework to generate a dense three-dimensional (3D) point cloud representation of each truck. Subsequently, PointNet – a deep representation learning algorithm – was adopted to train the classification model from reconstructed point clouds. The model utilizes low-level features extracted from the 3D point clouds, and detects key features associated with each truck class. Finally, model ensemble techniques were explored to reduce the model variance and further enhance the model performance. The developed model is capable of distinguishing passenger vehicles and 29 different truck body configurations. The average correct classification rate of the method on the test dataset is 94 percent for single and semi-trailer trucks, and 82 percent for single-unit vehicles with or without a trailer.
Analysis of Big Transportation Data for Better Infrastructure Management: A Case Study Using Very Large Weigh-in-Motion Data
Sami Demiroluk, AgileAssets, Inc.Show Abstract
Kaan Ozbay, New York University
Hani Nassif, Rutgers University
With the advance of technology, the size and availability of data increase for all sectors including transportation. Similarly, availability and size of Weigh-in-Motion (WIM) data increase every day, which help the professionals to monitor the trucks in the network and observe trends in the weights and configurations of the vehicles. However, truck monitoring does not occur in near real-time due to difficulties in data collection and processing the data especially if it is from large number of stations. Moreover, the “traditional” database systems are either too slow in analyzing these big datasets or charge substantial licensing fees per utilized CPU core. Apache Spark is an open-source big data analytics and distributed computing framework and it can offer a solution to aforementioned problems. In this study, the authors demonstrate the potential benefits of using Spark and cloud computing in analyzing WIM data using one of the largest datasets in the transportation literature containing over 400 million vehicle records from the statewide WIM sites in New Jersey. First, a set of queries are chosen, which are mostly essential for exploratory analyzing of WIM data. Utilizing another data source for overweight truck permits, more complex queries running on multiple datasets are also developed. Then, Spark is benchmarked against traditional databases using these queries. The performance of Spark on a computer cluster is also investigated with varying resource configurations. The results show that considerable computation time gains are possible in analyzing this large data using big data tools, especially on a computer cluster.
Quality Control of WIM-Based Live Load Data for Bridges
Anjan Ramesh Babu (firstname.lastname@example.org), Prime AE Group, IncShow Abstract
Olga Iatsko, Finley Engineering Group
Andrzej Nowak, Auburn University
J.Michael Stallings, Auburn University
Truckloads are a major factor affecting the service and fatigue life of bridges and pavements. The significant source of information about traffic load is the Weigh-in-Motion (WIM) database. Traffic data is used for the development of a bridge live load model for design and evaluation, prediction, and assessment of live loads. In pavement design, it is used for computing equivalent single-axle loads. However, poor quality of traffic data may lead to misinterpretation and incorrect estimation of load effects. Errors may occur due to WIM system malfunction, out-of-calibration, or irregular vehicle position on the sensor. If an error in recorded WIM data is not recognized and eliminated at an earlier stage, the quality of entire accumulated data is questionable. Therefore, it is essential to use a Quality Control (QC) procedure. This paper proposes a procedure to check the quality of traffic data and detect the root cause of questionable records. Inconsistency in recording due to communication failure, operational problems with the sensor and drift in calibration can be interpreted from this proposed procedure. It involves a completeness check, logical checks, and statistical checks. The limits of threshold parameters are established based on literature review and information on issued permits. The proposed QC procedure can verify the accuracy of unusual vehicle configurations that are categorized as “unclassified,” and it is demonstrated on the Alabama WIM data. A computer app, ALDOT_WIM_QC v1.0, was developed and implemented by ALDOT to evaluate the quality of WIM data and maintain the “health” of WIM systems.
Automatic Video Detection and Tracking of Pedestrians and Cyclists: Exploratory Feasibility Analysis
Deng Pan (email@example.com), George Washington UniversityShow Abstract
Samer Hamdar, George Washington University
Yufei Yuan, Technische Universiteit Delft
Victor Knoop, Delft University of Technology
Winnie Daamen, Delft University of Technology
Alireza Talebpour, University of Illinois, Urbana-Champaign
The objective of this paper is to automatically transform visual information extracted from videos into useful pedestrians’ and cyclists’ traffic measures. There are two main contribution associated with this objective: 1) Detecting and tracking the pedestrians and cyclists in real time. 2) Analyzing the traced trajectories to efficiently get useful information such as flows, densities and speeds. To evaluate the corresponding methodology, this paper compares the results with those obtained from an existing video detection and tracing exercise performed manually. The video data provided by the Delft University of Technology shows a mixed non-directional non-motorized traffic at shared space next to a transit station in Amsterdam, the Netherlands. YOLOv3 is used to detect the objects and Deep Sorting is used to trace locations across time. Comparing the automatically extracted data with the manually extracted data, the accuracy of the proposed method was found to be 100% for pedestrians (i.e. 100% of the pedestrians are detected) and 98.11% for cyclists. The average relative error speed is 0.040 m/s for cyclists and 0.056 m/s for pedestrians. According to the fundamental diagrams estimated based on the trajectory data, the cyclists’ free flow speed is found to be 5.19 meters per second and the collective jam density is 3.477 objects per square meter. The maximum flow recorded in the study area is 5.47 objects (pedestrian and cyclists) per second per meter. These measures are plausible and comparable to those reported in the literature. Keywords: Pedestrians, cyclists, mixed traffic, video detection, trajectories
Estimating Pedestrian Delay at Signalized Intersections Using Finite Mixture Modeling
Abolfazl Karimpour (firstname.lastname@example.org), University of ArizonaShow Abstract
Jason Anderson, Portland State University
Sirisha Kothuri, Portland State University
Yao-Jan Wu, University of Arizona
Estimating accurate pedestrian delay is necessary, as it is one of the fundamental multimodal signal performance measures. It has been widely shown that pedestrians’ level of frustration grows with the increase of pedestrian delay, which may cause pedestrians to violate the signals. However, for agencies seeking to use multimodal signal performances for signal operations, pedestrian delay is not always readily available. This study proposes a finite mixture modeling approach to estimate pedestrian delay as a function of specific covariates. The application of the proposed method was used to estimate delay at four intersections on a major corridor in Pima County, Arizona. The results showed the proposed method was able to capture and track the actual pedestrian delay fluctuation during the day with an average 10% mean-absolute-error. In addition, it was found that the proposed method is transferable and can be used as a network-wide delay estimation model for intersections with similar specifications. The application of the proposed method could provide agencies with a more reliable, robust, and yet accurate approach for estimating pedestrian delay at signalized intersections where the sensors are not available to collect pedestrian delay.
DISCLAIMER: All information shared in the TRB Annual Meeting Online Program is subject to change without notice. Changes, if necessary, will be updated in the Online Program and this page is the final authority on schedule information.