A process convolution model for crash count data on a network.

Crash data observed on a road network often exhibit spatial correlation due to unobserved effects with inherent spatial correlation following the structure of the road network. It is important to model this spatial correlation while accounting for the road network structure. In this study, we introduce the network process convolution (NPC) model. In this model, the spatial correlation among crash data is captured by a Gaussian Process (GP) approximated through a kernel convolution approach. The GP's covariance function is based on path distance computed between a limited set of knots and crash data points on the road network. The proposed model offers a straightforward approach for predicting crash frequency at unobserved locations where covariates are available, and for interpolating the GP values anywhere on the network. Inference procedure is performed following the Bayesian paradigm and is implemented in R-INLA, which offers an estimation procedure that is very efficient compared to Markov Chain Monte Carlo sampling algorithms. We fitted our model to synthetic data and to crash data from Ottawa, Canada. We compared the proposed approach with a proper Conditional Autoregressive (pCAR) model, and with Poisson Regression (PR) and Negative Binomial (NB) models without latent effects. The results of the study indicated that although the pCAR model has comparable fitting performance, the NPC model outperforms pCAR when the main goal is to predict unobserved locations of interest. The proposed model also offers lower mean absolute error rates for cross validated crash counts, latent variable values, fixed-effect coefficients, as well as shorter interval scores for singletons. The NPC provides a natural way to account for the road network structure when considering the inclusion of spatially structured latent random effects in the modelling of crash data. It also offers an improved predictive capability for crash data on a road network.

Evaluating safety-influencing factors at stop-controlled intersections using automated video analysis.

INTRODUCTION: Although stop signs are popular in North America, they have become controversial in cities like Montreal, Canada where they are often installed to reduce vehicular speeds and improve pedestrian safety despite limited evidence demonstrating their effectiveness. The purpose of this study is to evaluate the impact of stop-control configuration (and other features) on safety using statistical models and surrogate measures of safety (SMoS), namely vehicle speed, time-to-collision (TTC), and post-encroachment time (PET), while controlling for features of traffic, geometry, and built environment. METHODS: This project leverages high-resolution user trajectories extracted from video data collected for 100 intersections, 336 approaches, and 130,000 road users in Montreal to develop linear mixed-effects regression models to account for within-site and within-approach correlations. This research proposes the Intersection Exposure Group (IEG) indicator, an original method for classifying microscopic exposure of pedestrians and vehicles. RESULTS: Stop signs were associated with an average decrease in approach speed of 17.2 km/h and 20.1 km/h, at partially and fully stop-controlled respectively. Cyclist or pedestrian presence also significantly lower vehicle speeds. The proposed IEG measure was shown to successfully distinguish various types of pedestrian-vehicle interactions, allowing for the effect of each interaction type to vary in the model. CONCLUSIONS: The presence of stop signs significantly reduced approach speeds compared to uncontrolled approaches. Though several covariates were significantly related to TTC and PET for vehicle pairs, the models were unable to demonstrate a significant relationship between stop signs and vehicle-pedestrian interactions. Therefore, drawing conclusions regarding pedestrian safety is difficult. Practical Applications: As pedestrian safety is frequently used to justify new stop sign installations, this result has important policy implications. Policies implementing stop signs to reduce pedestrian crashes may be less effective than other interventions. Enforcement and education efforts, along with geometric design considerations, should accompany any changes in traffic control.

Massive GNSS data for road safety analysis: Comparing crash models for several Canadian cities and data sources.

Mobile sensors are a useful data source with applications in several transportation fields. Though cost of collection, transmission, and storage has limited studies on driving data and safety, this can be overcome through usage-based insurance (UBI). In UBI programs, drivers are monitored, and their premiums are adjusted based on driver-level surrogate safety measures (SSMs) related to exposure and driving style. Contextual link-level SSMs (volume, speed, or density) could further improve discount calibration. This study quantifies relationships between contextual SSMs and crashes and includes the validation of previous results (correlations between SSMs and crashes and statistical models estimated using smartphone-collected data from Quebec City) and the comparison of three Canadian cities (using UBI data from Quebec City, Montreal, and Ottawa). Extracted SSMs were compared to large volumes of historical crash frequency data using Spearman's Rank Correlation Coefficient and then implemented into spatial Bayesian crash models. Results from the UBI data generally matched those from the previous study, with observed correlations mirroring previous results in direction (braking, congestion, and speed variation are positively associated with crash frequency while mean speed is negatively associated) while correlation strength was slightly higher. Furthermore, these results were consistent between cities. For the crash modelling, repeatability of previous results in Quebec City was moderately good for the UBI data. Importantly for large-scale implementation, models estimated using UBI data were largely consistent between cities. This work provides an important contribution to the existing literature, clearly demonstrating how contextual safety measures could be applied to benefit UBI practices.

Pedestrian safety at signalized intersections: Modelling spatial effects of exposure, geometry and signalization on a large urban network.

Intersections represent the most dangerous sites in the road network for pedestrians: not only is modal separation often impossible, but elements of geometry, traffic control, and built environment further exacerbate crash risk. Evaluating the safety impact of intersection features requires methods to quantify relationships between different factors and pedestrian injuries. The purpose of this paper is to model the effects of exposure, geometry, and signalization on pedestrian injuries at urban signalized intersections using a Full Bayes spatial Poisson Log-Normal model that accounts for unobserved heterogeneity and spatial correlation. Using the Integrated Nested Laplace Approximation (INLA) technique, this work leverages a rich database of geometric and signalization variables for 1864 intersections in Montreal, Quebec. To collect exposure data, short-term pedestrian and vehicle counts were extrapolated to AADT using developed expansion factors. Results of the model confirmed the positive relationship between pedestrian and vehicle volumes and pedestrian injuries. Curb extensions, raised medians, and exclusive left turn lanes were all found to reduce pedestrian injuries, while the total number of lanes and the number of commercial entrances were found to increase them. Pedestrian priority phases reduced injuries while the green straight arrow increased injuries. Lastly, the posterior expected number of crashes was used to identify hotspots. The proposed ranking criteria identified many intersections close to the city centre where the expected number of crashes is highest and intersections along arterials with lower pedestrian volumes where individual pedestrian risk is elevated. Understanding the effects of intersection geometry and pedestrian signalization will aid in ensuring the safety of pedestrians at signalized intersections.

Network screening for large urban road networks: Using GPS data and surrogate measures to model crash frequency and severity.

Crash frequency and injury severity are independent dimensions defining crash risk in road safety management and network screening. Traditional screening techniques model crashes using regression and historical crash data, making them intrinsically reactive. In response, surrogate measures of safety have become a popular proactive alternative. The purpose of this paper is to develop models for crash frequency and severity incorporating GPS-derived surrogate safety measures (SSMs) as predictive variables. SSMs based on vehicle manoeuvres and traffic flow were extracted from data collected in Quebec City. The mixed multivariate outcome is estimated using two models; a Full Bayes Spatial Negative Binomial model for crash frequency estimated using the Integrated Nested Laplace Approximation approach and a fractional Multinomial Logit model for crash severity. Model outcomes are combined to generate posterior expected crash frequency at each severity level and rank sites based on crash cost. The crash frequency model was accurate at the network scale, with the majority of proposed SSMs statistically significant at 95% confidence and the direction of their effect generally consistent with previous research. In the crash severity model, fewer variables were significant, yet the direction of the effect of all significant variables was again consistent with previous research. Correlations between rankings predicted by the mixed multivariate model and by the crash data were adequate for intersections (0.46) but were poorer for links (0.25). The ability to prioritize sites based on GPS data and SSMs rather than historical crash data represents a substantial contribution to the field of road safety.

Modeling when and where a secondary accident occurs.

The occurrence of secondary accidents leads to traffic congestion and road safety issues. Secondary accident prevention has become a major consideration in traffic incident management. This paper investigates the location and time of a potential secondary accident after the occurrence of an initial traffic accident. With accident data and traffic loop data collected over three years from California interstate freeways, a shock wave-based method was introduced to identify secondary accidents. A linear regression model and two machine learning algorithms, including a back-propagation neural network (BPNN) and a least squares support vector machine (LSSVM), were implemented to explore the distance and time gap between the initial and secondary accidents using inputs of crash severity, violation category, weather condition, tow away, road surface condition, lighting, parties involved, traffic volume, duration, and shock wave speed generated by the primary accident. From the results, the linear regression model was inadequate in describing the effect of most variables and its goodness-of-fit and accuracy in prediction was relatively poor. In the training programs, the BPNN and LSSVM demonstrated adequate goodness-of-fit, though the BPNN was superior with a higher CORR and lower MSE. The BPNN model also outperformed the LSSVM in time prediction, while both failed to provide adequate distance prediction. Therefore, the BPNN model could be used to forecast the time gap between initial and secondary accidents, which could be used by decision makers and incident management agencies to prevent or reduce secondary collisions.

Surrogate safety and network screening: Modelling crash frequency using GPS travel data and latent Gaussian Spatial Models.

Improving road safety requires accurate network screening methods to identify and prioritize sites in order to maximize the effectiveness of implemented countermeasures. In screening, hotspots are commonly identified using statistical models and ranking criteria derived from observed crash data. However, collision databases are subject to errors, omissions, and underreporting. More importantly, crash-based methods are reactive and require years of crash data. With the arrival of new technologies including Global Positioning System (GPS) trajectory data, proactive surrogate safety methods have gained popularity as an alternative approach for screening. GPS-enabled smartphones can collect reliable and spatio-temporally rich driving data from regular drivers using an inexpensive, simple, and user-friendly tool. However, few studies to date have analyzed large volumes of smartphone GPS data and considered surrogate-safety modelling techniques for network screening. The purpose of this paper is to propose a surrogate safety screening approach based on smartphone GPS data and a Full Bayesian modelling framework. After processing crash data and GPS data collected in Quebec City, Canada, several surrogate safety measures (SSMs), including vehicle manoeuvres (hard braking) and measures of traffic flow (congestion, average speed, and speed variation), were extracted. Then, spatial crash frequency models incorporating the extracted SSMs were proposed and validated. A Latent Gaussian Spatial Model was estimated using the Integrated Nested Laplace Approximation (INLA) technique. While the INLA Negative Binomial models outperformed alternative models, incorporating spatial correlations provided the greatest improvement in model fit. Relationships between SSMs and crash frequency established in previous studies were generally supported by the modelling results. For example, hard braking, congestion, and speed variation were all positively linked to crash counts at the intersection level. Network screening based on SSMs presents a substantial contribution to the field of road safety and works towards the elimination of crash data in evaluation and monitoring.

Vehicle manoeuvers as surrogate safety measures: Extracting data from the gps-enabled smartphones of regular drivers.

Network screening is a key element in identifying and prioritizing hazardous sites for engineering treatment. Traditional screening methods have used observed crash frequency or severity ranking criteria and statistical modelling approaches, despite the fact that crash-based methods are reactive. Alternatively, surrogate safety measures (SSMs) have become popular, making use of new data sources including video and, more rarely, GPS data. The purpose of this study is to examine vehicle manoeuvres of braking and accelerating extracted from a large quantity of GPS data collected using the smartphones of regular drivers, and to explore their potential as SSMs through correlation with historical collision frequency and severity across different facility types. GPS travel data was collected in Quebec City, Canada in 2014. The sample for this study contained over 4000 drivers and 21,000 trips. Hard braking (HBEs) and accelerating events (HAEs) were extracted and compared to historical crash data using Spearman's correlation coefficient and pairwise Kolmogorov-Smirnov tests. Both manoeuvres were shown to be positively correlated with crash frequency at the link and intersection levels, though correlations were much stronger when considering intersections. Locations with more braking and accelerating also tend to have more collisions. Concerning severity, higher numbers of vehicle manoeuvres were also related to increased collision severity, though this relationship was not always statistically significant. The inclusion of severity testing, which is an independent dimension of safety, represents a substantial contribution to the existing literature. Future work will focus on developing a network screening model that incorporates these SSMs.

The impact of vehicle moving violations and freeway traffic flow on crash risk: An application of plugin development for microsimulation.

This paper presents the use of the Aimsun microsimulation program to simulate vehicle violating behaviors and observe their impact on road traffic crash risk. Plugins for violations of speeding, slow driving, and abrupt stopping were developed using Aimsun's API and SDK module. A safety analysis plugin for investigating probability of rear-end collisions was developed, and a method for analyzing collision risk is proposed. A Fuzzy C-mean Clustering algorithm was developed to identify high risk states in different road segments over time. Results of a simulation experiment based on the G15 Expressway in Shanghai showed that abrupt stopping had the greatest impact on increasing collision risk, and the impact of violations increased with traffic volume. The methodology allows for the evaluation and monitoring of risks, alerting of road hazards, and identification of hotspots, and could be applied to the operations of existing facilities or planning of future ones.

Predicting drowsy driving in real-time situations: Using an advanced driving simulator, accelerated failure time model, and virtual location-based services.

This paper aims to both identify the factors affecting driver drowsiness and to develop a real-time drowsy driving probability model based on virtual Location-Based Services (LBS) data obtained using a driving simulator. A driving simulation experiment was designed and conducted using 32 participant drivers. Collected data included the continuous driving time before detection of drowsiness and virtual LBS data related to temperature, time of day, lane width, average travel speed, driving time in heavy traffic, and driving time on different roadway types. Demographic information, such as nap habit, age, gender, and driving experience was also collected through questionnaires distributed to the participants. An Accelerated Failure Time (AFT) model was developed to estimate the driving time before detection of drowsiness. The results of the AFT model showed driving time before drowsiness was longer during the day than at night, and was longer at lower temperatures. Additionally, drivers who identified as having a nap habit were more vulnerable to drowsiness. Generally, higher average travel speeds were correlated to a higher risk of drowsy driving, as were longer periods of low-speed driving in traffic jam conditions. Considering different road types, drivers felt drowsy more quickly on freeways compared to other facilities. The proposed model provides a better understanding of how driver drowsiness is influenced by different environmental and demographic factors. The model can be used to provide real-time data for the LBS-based drowsy driving warning system, improving past methods based only on a fixed driving.

Investigating the gender differences on bicycle-vehicle conflicts at urban intersections using an ordered logit methodology.

In the literature, a crash-based modeling approach has long been used to evaluate the factors that contribute to cyclist injury risk at intersections. However, this approach has been criticized as crashes are required to occur before contributing factors can be identified and countermeasures can be implemented. Moreover, human factors related to dangerous behaviors are difficult to evaluate using crash-based methods. As an alternative, surrogate safety measures have been developed to address the issue of reliance on crash data. Despite recent developments, few methodologies and little empirical evidence exist on bicycle-vehicle interactions at intersections using video-based data and statistical analyses to identify associated factors. This study investigates bicycle-vehicle conflict severity and evaluates the impact of different factors, including gender, on cyclist risk at urban intersections with cycle tracks. A segmented ordered logit model is used to evaluate post-encroachment time between cyclists and vehicles. Video data was collected at seven intersections in Montreal, Canada. Road user trajectories were automatically extracted, classified, and filtered using a computer vision software to yield 1514 interactions. The discrete choice variable was generated by dividing post-encroachment time into normal interactions, conflicts, and dangerous conflicts. Independent variables reflecting attributes of the cyclist, vehicle, and environment were extracted either automatically or manually. Results indicated that an ordered model is appropriate for analyzing traffic conflicts and identifying key factors. Furthermore, exogenous segmentation was beneficial in comparing different segments of the population within a single model. Male cyclists, with all else being equal, were less likely than female cyclists to be involved in conflicts and dangerous conflicts at the studied intersections. Bicycle and vehicle speed, along with the time of the conflict relative to the red light phase, were other significant factors in conflict severity. These results will contribute to and further the understanding of gender differences in cycling within North America.