Transportation Safety

One of urbanMITS's research areas is centered on transportation safety. Our long-term goal is dedicated to Vision Zero: zero deaths, zero injuries, and zero fear of traffic. It is concerned with the study of safety issues among different traffic entities such as pedestrian, bicyclists, vehicles, etc. This involves effectively tap knowledge in different disciplines to understand the situation, trend, possible factors and to provide support for executing research/field projects and treatments that help account for the traffic safety issues in local NYC community and beyond. The researchers from the NYU urbanMITS are actively exploring relevant data, countermeasures and policy that help address the safety issues. A number of theories, analytical methodologies and techniques for enlarging our understanding of traffic safety are being developed. Extensive data sources such as historical crashes, real-time traffic surrogate safety measures, user behavior data as well as sensor data are explored.

Selected Research Topics

Secondary Crash Modeling and Analysis

Mining the Characteristics of Secondary Crashes on Highways
 Author: Hong Yang, Bekir Bartin, Kaan Ozbay

 Journal: Journal of Transportation Engineering, DOI: 10.1061/(ASCE)TE.1943-5436.0000646

 Abstract: The prevention of secondary crashes is a high priority task in traffic incident management. However, the limited knowledge regarding the nature of secondary crashes largely impeded the development of established countermeasures. The primary goal of this paper is to improve the literature's understanding of secondary crashes. This goal is achieved in two steps: first, with an analysis framework that accurately identifies secondary crashes by integrating rich traffic-sensor data with statewide-crash data and, second, by carefully investigating the characteristics of these identified secondary crashes. To that end, secondary crashes within a 27-mile section of a major highway in New Jersey were mined using the developed analysis framework, and a thorough examination of their characteristics has been performed. Empirical findings on the frequency of secondary crashes, their spatio-temporal distributions, clearance time, crash type, severity, and major contributing factors have been highlighted. Taken together, these preliminary results could potentially help transportation agencies make more informed decisions on mitigating secondary crashes and improve their incident management operations. To complement the results, further in-depth investigations using more high-resolution sensor data and high-quality incident records are suggested.

Download PDF  Download Slides
Development of an On-line Scalable Approach for Identifying Secondary Crashes
 Author: Hong Yang, Kaan Ozbay, Ender Faruk Morgul, Bekir Bartin, Kun Xie

 Journal: Journal of Transportation Research Record, (Freeway Operations Committee (AHB20))

 Conference: Transportation Research Board 93rd Annual Meeting, Washington D. C., January 12-16, 2014

 Abstract: Secondary crashes are one of the most critical incidents occurring on highways. They can induce extra traffic delays and affect highway safety performance. Transportation agencies are interested in understanding the mechanism of the secondary crash occurrence and implementing appropriate countermeasures. However, there is no well-established procedure to identify secondary crashes, which in turn impedes the possibility of investigating the underlying mechanism of their occurrence. This study intend to develop an on-line scalable approach to help identify secondary crashes for a large number of highways that have insufficient traffic surveillance units collecting continuous traffic data required to classify secondary crash accurately. The developed approach consists of two major components: (a) acquisition of open source traffic data and (b) identification of secondary crashes through the use of these data. Unlike existing approaches based on static thresholds, queuing models or infrastructure-based sensor data, the developed approach takes advantage of various open sources data to identify traffic conditions in the presence of incidents. In this study, we propose to develop virtual sensors collecting traffic data from private traffic information providers such as Bing Maps, Google Maps and MapQuest. The availability of such data greatly expands our ability to cover more highways without installing infrastructure sensors. The virtual sensor output provides the basic input to run the developed automatic identification algorithm for identifying secondary crashes. The algorithm is described in a step-by-step manner to provide a readily deployable approach for transportation agencies interested in identifying secondary crashes on their highway networks.

Download PDF  Download Slides
Assessing the Risk of Secondary Crashes on Freeways
 Author: Hong Yang, Kaan Ozbay, Kun Xie

 Journal: Journal of Safety Research,

 Conference: Road Safety and Simulation International Conference (RSS2013), Rome, Italy, October 22-25, 2013

 Abstract: Introduction: The occurrence of secondary crashes is one of the critical yet understudied highway safety issues. Induced by the primary crashes, the occurrence of secondary crashes does not only increase traffic delays but also the risk of inducing additional incidents. Many highway agencies are highly interested in the implementation of safety countermeasures to reduce this type of crashes. However, due to the limited understanding of the key contributing factors, they face a great challenge for determining the most appropriate countermeasures.Method: To bridge this gap, this study makes important contributions to the existing literature of secondary incidents by developing a novel methodology to assess the risk of having secondary crashes on highways. The proposed methodology consists of two major components, namely: (a) accurate identification of secondary crashes and (b) statistically robust assessment of causal effects of contributing factors. The first component is concerned with the development of an improved identification approach for secondary accidents that relies on the rich traffic information obtained from traffic sensors. The second component of the proposed methodology is aimed at understanding the key mechanisms that are hypothesized to cause secondary crashes through the use of a modified logistic regression model that can efficiently deal with relatively rare events such as secondary incidents. The feasibility and improved performance of using the proposed methodology are tested using real-world crash and traffic flow data. Results: The risk of inducing secondary crashes after the occurrence of individual primary crashes under different circumstances is studied by employing the estimated regression model. Marginal effect of each factor on the risk of secondary crashes is also quantified and important contributing factors are highlighted and discussed. Practical applications: Massive sensor data can be used to support the identification of secondary crashes. The occurrence mechanism of these secondary crashes can be investigate by the proposed model. Understanding the mechanism helps deploy appropriate countermeasures to mitigate or prevent the secondary crashes.

Download PDF  Download Slides
Identifying Secondary Crashes on Freeways Using Sensor Data
 Author: Hong Yang, Bekir Bartin, and Kaan Ozbay

 Journal: Journal of Transportation Research Record, (In Press, 2013)

 Conference: In Transportation Research Board 92nd Annual Meeting Compendium of Papers CD-ROM, Washington D. C., January 13-17

 Abstract: Non-recurring traffic incidents such as motor vehicle crashes increase not only travel delays but also the risk of secondary crashes. Secondary crashes can cause additional traffic delays, and reduce safety. In order to implement effective countermeasures to prevent and/or reduce secondary crashes, first their characteristics should be investigated. However, the related research has been limited largely due to the lack of detailed incident and traffic data necessary to first identify the secondary crashes. Existing approaches such as static methods employed to identify secondary crashes cannot fully capture potential secondary crashes due to fixed spatio-temporal identification criteria. Improved approaches are needed to accurately categorize secondary crashes for further analysis. Therefore, this paper attempts to develop an enhanced approach for identifying secondary crashes using the existing crash database and archived traffic data from highway sensors. The proposed method is threefold: (a) defining secondary crashes; (b) examining the impact range of primary crashes that possibly relate to secondary crashes; and (c) identifying secondary crashes. The proposed methodology establishes a practical framework for mining secondary crashes from existing sensor data and crash records. A case study is performed on a 27-mile segment of a major highway in New Jersey to illustrate the performance of the proposed approach. The results show that the proposed method provides a more reliable and efficient categorization of secondary crashes than commonly used approaches.

Download PDF  Download Slides
Investigating the Characteristics of Secondary Crashes on Freeways
 Author: Hong Yang, Bekir Bartin, and Kaan Ozbay

 Conference: In Transportation Research Board 92nd Annual Meeting Compendium of Papers CD-ROM, Washington D. C., January 13-17

 Abstract: Prevention of secondary crashes is one of the priorities in traffic incident management. However, limited information on secondary crashes has largely impeded the selection of appropriate countermeasures. The primary goal of this paper is to improve the understanding of secondary crashes, which is achieved by two major steps. First, an analysis framework is developed to accurately identify secondary crashes by integrating rich traffic sensor data with the statewide crash data sets. Second, the characteristics of the identified secondary crashes are investigated in detail. Secondary crashes that occurred on a 27-mile section of a major highway in New Jersey were mined using the proposed analysis framework. A thorough examination of their characteristics was then performed. Empirical findings on the frequency of secondary crashes, their spatio-temporal distributions, clearance time, crash type, severity, and major contributing factors were highlighted. These preliminary results can help transportation agencies make more informed decisions on mitigating secondary crashes and improve their incident management operations. To complement the results, further in-depth investigations based on more high-resolution sensor data and high-quality incident records are suggested.

Download PDF  Download Slides

Work Zone Safety Impact

Estimating the Impact of Work Zones on Highway Safety
 Author: Ozgur Ozturk, Kaan Ozbay, Hong Yang

 Conference: Transportation Research Board 93rd Annual Meeting, Washington D. C., January 12-16, 2014,

 Abstract: Investigating the operational and safety impacts of work zones on traffic are of great interest to transportation agencies. Despite the increasing attention in modeling work zone crash frequency, only a limited number of studies directly examined the change in crash rates under work zone conditions versus the corresponding non-work zone conditions. The main research question is thus whether or not a given roadway experiences increased number of crashes in the presence of a work zone. Another important goal of this study is to address the main research question using a large number of relatively long-term work zones. This is a challenging task because a major effort for building a comprehensive data set of a large number of work zones is needed. To answer this important question, an integrated data set is created by aggregating information from multiple data sources for sixty long-term work zones. Then descriptive analysis is performed to examine the characteristics of work zone crashes. Preliminary analysis results show that the crash rate increased by 24.4 percent under the work zone condition. To further statistically quantify the impact of work zones on traffic safety, negative binomial regression models are developed to identify relationship between a set of covariates and the crash counts aggregated by time and crash severity. An indicative variable is incorporated into the model to represent the work zone and non-work zone conditions for the studied sites. Overall, the descriptive analysis and modeling results suggest that the presence of work zones significantly increases the risk of crashes on roads.

Download PDF  Download Slides
Investigating Motorists' Behaviors in Response to Supplementary Traffic Control Devices at Land Surveying Work Sites
 Author: Hong Yang, Bekir Bartin, Kaan Ozbay, and S. Chien

 Journal: Traffic Injury Prevention, DOI: http://dx.doi.org/10.1080/15389588.2013.823165

 Abstract:

Download PDF  Download Slides
Modeling Work Zone Crash Frequency by Quantifying Measurement Errors in Work Zone Length
 Author: Hong Yang, Kaan Ozbay, Ozgur Ozturk, and M. Yildirimoglu

 Journal: Accident Analysis & Prevention, Vol. 55, June 2013, pp. 192-201

 Abstract: Work zones are temporary traffic control zones that can potentially cause safety problems. Maintaining safety, while implementing necessary changes on roadways, is an important challenge traffic engineers and researchers have to confront. In this study, the risk factors in work zone safety evaluation were identified through the estimation of a crash frequency (CF) model. Measurement errors in explanatory variables of a CF model can lead to unreliable estimates of certain parameters. Among these, work zone length raises a major concern in this analysis because it may change as the construction schedule progresses generally without being properly documented. This paper proposes an improved modeling and estimation approach that involves the use of a measurement error (ME) model integrated with the traditional negative binomial (NB) model. The proposed approach was compared with the traditional NB approach. Both models were estimated using a large dataset that consists of 60 work zones in New Jersey. Results showed that the proposed improved approach outperformed the traditional approach in terms of goodness-of-fit statistics. Moreover it is shown that the use of the traditional NB approach in this context can lead to the overestimation of the effect of work zone length on the crash occurrence.

Download PDF  Download Slides
Work Zone Safety Analysis and Modeling: A State-of-the-Art Review
 Author: Hong Yang, Ozgur Ozturk, Kaan Ozbay, Kun Xie

 Conference: Transportation Research Board 93rd Annual Meeting, Washington D. C., January 12-16, 2014,

 Abstract: Prevention of work zone crashes is one of the top priorities for transportation agencies. To make more informed decisions on initiating appropriate programs and countermeasures, more precise information on the underlying mechanisms of work zone crashes is needed. Considerable research effort has been directed to examine work zone crash characteristics and possible causal factors in the last five decades. This study is designed to provide a thorough review of existing research focused on work zone crash related analysis and modeling. Literature from multiple sources is carefully combed to determine about the state-of-knowledge related to the work zone safety problem. Experience and lessons learned from these studies are presented to highlight critical gaps in the knowledge of the safety of work zones. Future challenges are also discussed to address some missing information needs related to work zone safety analysis and modeling.

Download PDF  Download Slides
Modeling Driver Injury Severity of Single-Vehicle Crashes in Freeway Construction Work Zones
 Author: Hong Yang, Ozgur Ozturk, Kaan Ozbay, and Bekir Bartin

 Conference: Selected Proceeding of 13th World Conference on Transport Research Society, Rio de Janeiro, Brazil, July 15-18, 2013

 Abstract:

Download PDF  Download Slides
Crash Frequency Modeling for Highway Construction Zones
 Author: Ozgur Ozturk, Kaan Ozbay, Hong Yang, and Bekir Bartin

 Conference: In Transportation Research Board 92nd Annual Meeting Compendium of Papers CD-ROM, Washington D. C., January 13-17

 Abstract: Throughout the country, work zone safety issues have received considerable attention in recent years due to increasing work zone crashes along numerous highway renovation and reconstruction projects. In general, previous studies were able to consider limited number of contributing factors mainly due to the limitations in data availability. The main goal of this paper is to remedy this major problem related to data availability and estimate improved models using data from multiple sources. Work zone project drawings, crash database and straight line diagrams are used to create an integrated work zone safety database. Work zone crash data is plotted in time and space to validate, locate and adjust work zone related information. The negative binomial regression approach is used as the appropriate model to predict crash frequency within these work zones. Traffic volume is adjusted for daytime and night time conditions in terms of hourly distribution of daily traffic. The duration-based and period-based models are also developed to address relationship between potential factors and to predict crash frequency on work zones in terms of property damage only (PDO) and injury crashes. Compared with previous frequency models, additional parameters such as number of lanes closed and speed reduction are used. These additional factors identified as significant can help traffic engineers to further improve safety of work zone projects.

Download PDF  Download Slides
Modeling Crash Risk of Highway Work Zones with Relatively Short Durations
 Author: Hong Yang, Kaan Ozbay, Kun Xie, Bekir Bartin

 Conference: 94th TRB Annual Conference (CD-ROM), Washington, D.C., January, 2015., (Work Zone Traffic Control Committee (AHB55))

 Abstract: Highway work zones greatly affect operational and safety performance of traffic. Existing studies have primarily focused on exploring safety issues of long-term work zones whereas safety issues associated with a large number of relatively shorter duration work zones are seldom examined. Thus, this paper aims to present an investigation of traffic safety in these work zones with relatively short duration. Considering low frequency of crashes due to this type of short duration work zones with respect to its crash condition and non-crash condition, a rare event logistic regression model is developed to explore the causal relationship between a set of contributing factors and the crash risk. The proposed model accounts for the imbalance issues of events (crashes) vs. non-event (non-crash) conditions in modeling low frequency crash occurrences. In addition, the model uses actual traffic data to reduce the bias of using aggregated exposure data (i.e, averaged traffic volume over a day, month, year, etc.). The modeling results based on a case study show that the work zone length, traffic volume and lane closure are positively associated with the crash risk in those work zones with short duration. The proposed model with specific model corrections and actual input data is found to improve the depiction of the relationship between these factors and the crash risk based on the comparison of the area under the ROC (Receiver Operating Characteristics) curves of different models.

Download PDF  Download Slides
Investigating the Impact of Work Zones on Crash Severity by Comparative Analysis
 Author: Ozgur Ozturk, Kaan Ozbay, Hong Yang

 Conference: 94th TRB Annual Conference (CD-ROM), Washington, D.C., January, 2015. , (Safety Data, Analysis and Evaluation Committee (ANB20))

 Abstract: Work zone safety has received much attention in recent years due to numerous highway construction projects that have resulted in many work zone crashes. To minimize the impact of work zones on roadway safety, the contributing potential factors that influence these risks need to be investigated. This can be done by identifying the possible causal factors in terms of crash severity and implementing countermeasures to ensure the motorists safety.

Download PDF  Download Slides

Development and Using Surrogate Safety Measures

Derivation and Validation of New Simulation-Based Surrogate Safety Measure
 Author: Kaan Ozbay, Hong Yang, Bekir Bartin, and S. Mudigonda

 Journal: Journal of Transportation Research Record, No. 2083, pp. 105-113

 Abstract: 13

Download PDF  Download Slides

Highway and Toll Plaza Safety Assessment

Effects of the Open Road Tolling on Safety Performance of Freeway Mainline Toll Plazas
 Author: Hong Yang, Kaan Ozbay, and Bekir Bartin

 Journal: Journal of Transportation Research Record, No. 2324, pp. 101-109

 Abstract: Advances of Intelligent Transportation Systems (ITS) technologies promoted the implementation of open road tolling (ORT) on tolled freeways worldwide. This new tolling solution converts existing barrier tollbooths to express lanes capable of collecting tolls at high-speeds. ORT has demonstrated numerous benefits in reducing traffic congestion and air pollution. However, effects of ORT on safety are still not clear, as most of ORT systems have only been operated for a relatively short period of time. Therefore, this study aims to evaluate the safety impacts of ORT by studying locations where such tolling solution was recently deployed on the Garden State Parkway in New Jersey. Multiple-year crash data at the toll plazas before-after the implementation of the ORT systems were used for analysis. Full Bayes methodology is employed to estimate crash frequency models as a function of traffic and toll plaza configurations. These models were used to estimate the crash frequency assuming that the ORT systems were not installed. Then, these estimations were compared with the observed number of crashes occurred after the deployment of the ORT systems. Individual comparisons show that crash reductions are observed at most of the toll plazas. The overall comparison shows that crashes at locations where ORT systems were deployed are decreased by about 24 percent after deployment of these systems. It can thus be concluded that the use of ORT is a beneficial solution towards improving toll road safety. From an implementation point of view, the analyses results indicate that special attention should be paid to operational elements such as signage, diversion and merge designs of the ORT systems.

Download PDF  Download Slides
Effect of Removing Freeway Mainline Barrier Toll Plazas on Safety
 Author: Hong Yang, Kaan Ozbay, Bekir Bartin, and Ozgur Ozturk

 Conference: In Transportation Research Board 92nd Annual Meeting Compendium of Papers CD-ROM, Washington D. C., January 13-17

 Abstract: Toll plaza safety is a critical issue. Toll plazas induce motor vehicle crashes and also put workers such as toll collectors at risk. Therefore, enhancing safety at a toll plaza is crucial to improving safety on tolled roadways. This study aims to evaluate the safety effect of removing mainline barrier toll plazas on highways using Empirical Bayesian (EB) methodology. Recent removals of barrier toll plaza on the Garden State Parkway in New Jersey were used as a case study. Multiple-year traffic and crash data before and after the removals of the barrier toll plazas were used for analysis. Toll plaza crash frequency models as a function of traffic flow and other factors were developed, with the modeling results suggesting that there is a nonlinear relationship between toll plaza crash occurrences and both traffic flow as well as toll booth configurations. The EB approach is also used to predict crash frequency assuming that the barrier toll booths were not removed. These EB-based estimates were compared with the observed number of crashes after the removals of the toll plazas. Individual comparisons show reductions in crash frequency at almost all of the toll plazas and an estimated reduction of 47.2 percent overall at all toll plazas due to the removal of the barrier toll booths. The estimated crash cost was reduced by 43.2 percent. These estimated reductions demonstrate that the removal of barrier toll plazas is a very beneficial step towards improving safety of toll roads.

Download PDF  Download Slides

Intersection Safety Research

Crash Frequency Modeling for Signalized Intersections in a High-Density Urban Road Network
 Author: Kun Xie, X. Wang, Kaan Ozbay, and Hong Yang

 Journal: Analytic Methods in Accident Research,

 Conference: Road Safety and Simulation International Conference (RSS2013), Rome, Italy, October 22-25, 2013

 Abstract: Conventional crash frequency models rely on an assumption of independence among observed crashes. However, this assumption is frequently proved false by spatially related crash observations, particularly for intersection crashes observed in high-density road networks. Crash frequency models that ignore the hierarchy and spatial correlation of closely spaced intersections can lead to biased estimations. As a follow-up to our previous paper (Xie et al., 2013), this study aims to address this issue by introducing an improved crash frequency model. Data for 195 signalized intersections along 22 corridors in the urban areas of Shanghai was collected. Moran?s I statistic of the crash data confirmed the spatial dependence of crash occurrence among the neighboring intersections. Moreover, Lagrange Multiplier test was performed and it suggested that the spatial dependence should be captured in the model error term. A hierarchical model incorporating a conditional autoregressive (CAR) effect term for the spatial correlation was developed in the Bayesian framework. A deviance information criterion (DIC) and cross-validation test were used for model selection and comparison. The results showed that the proposed model outperformed traditional models in terms of the overall goodness of fit and predictive performance. In addition, the significance of the corridor-specific random effect and CAR effect revealed strong evidence for the presence of heterogeneity across corridors and spatial correlation among intersections.

Download PDF  Download Slides

Non-Motorized Safety Research

Enhancing the Quality of Infrared-Based Automatic Pedestrian Sensor Data by Nonparametric Statistical Method
 Author: Hong Yang, Kaan Ozbay, and Bekir Bartin

 Journal: Journal of Transportation Research Record, No. 2264, pp. 11-17

 Conference: Transportation Research Board 90th Annual Meeting

 Abstract: With the advent of Intelligent Transportation Systems in the last several decades, sensors are being extensively used to detect and count vehicle movements. On the other hand, the use of similar sensing technologies to detect pedestrian movements is relatively new. Pedestrian counts are essential for decision-making for pedestrian facility planning, signal timing, and pedestrian safety modeling. Conventional methods such as manual counting and videotaping can hardly satisfy the requirements of long-term pedestrian data collection programs. Fortunately, advances in sensing technologies have increased the ability of automating pedestrian data collection using infrared sensors. However, data quality of the infrared sensors is still a problem because several field studies showed that this type of sensors do not always perform perfectly. Field tests conducted by this study and by other research teams show that infrared sensors usually count significantly less than the actual number of pedestrians. Thus there are needs to enhance the data quality of infrared sensors. This paper proposes a nonparametric statistical method to calibrate raw sensor data to achieve this goal. Instead of using regression-based approaches that are traditionally preferred by traffic engineers, a bivariate bootstrap sampling procedure is used to obtain correction factors for new counts. Two case studies are used to test the validation of the proposed calibration procedure. Test results show that the proposed procedure can improve the sensor data quality in terms of reducing the discrepancy between sensor counts and ground truth data. The transferability of the calibration procedure is also verified through the case studies.

Download PDF  Download Slides
Investigating the Performance of Automatic Counting Sensors for Pedestrian Traffic Data Collection
 Author: Hong Yang, Kaan Ozbay, and Bekir Bartin

 Conference: In Selected Proceedings of the 12th World Conference on Transport Research Society, ISBN 978-989-96986-1-1 (Editors: J. Viegas and R. Macrio), Lisbon, Portugal, July 11-15, Paper ID C5-02147

 Abstract: 26

Download PDF  Download Slides