Analytic Methods in Accident Research最新文献

Real-time prediction of crash risk is an effective method for enhancing traffic safety, but it is not fully explored in freeway tunnels. A two-stage deep learning modeling framework comprising a preliminary exploration stage and a prediction and analysis stage is proposed for real-time crash risk prediction in freeway tunnels. A random parameters logit model with heterogeneity in means and variances is used in the preliminary exploration stage to investigate the unobserved heterogeneity and influence mechanism of precursors on real-time crash risk. In the prediction and analysis stage, a random deep and cross network model considering feature interactions and unobserved heterogeneities is developed to predict and analyze real-time crash risk, which is interpreted by the shapley additive explanations approach. The multi-source fusion dataset, collected from the Caltrans performance measurement system and the weather information website, is used to validate the proposed framework for exploring real-time crash risk in freeway tunnels. Results reveal that: (1) the random parameters logit model with heterogeneity in means and variances outperforms the traditional logit model in terms of the model fitting, providing a reference for deep learning modeling that may be able to improve model performance by addressing heterogeneity; (2) the important crash precursors such as the average difference in speed between detectors of tunnel entrance and exit are discovered based on the marginal effect analysis of the random parameters logit model with heterogeneity in means and variances; (3) the random deep and cross network model yields the best prediction performance compared to its counterparts (some other data-driven models), demonstrating the superior performance of deep learning models for real-time risk prediction tasks. It also indicates that considering feature interaction and heterogeneity in deep learning modeling can improve prediction performance; and (4) the important precursors found in the random deep and cross network model using the shapley additive explanations approach are close to those discovered in the statistical model, indicating that the proposed deep learning model can capture the similar effects of precursors as the statistical models, and the precursor interactions and heterogeneities also can be observed by the shapley additive explanations approach.

Using traffic conflict-based extreme value theory (EVT) models to quantify real-time crash-risk of road facilities is a promising direction for developing proactive traffic safety management strategies. Existing EVT real-time crash-risk analysis studies have only focused on using block maxima models. This study proposes a dynamic Bayesian hierarchical peak over threshold modeling approach to estimate real-time crash-risk based on traffic conflicts. The proposed approach combines quantile regression, dynamic updating approach, Bayesian hierarchical structure, and the peak over threshold method to generate time-varying generalized Pareto distributions to derive real-time crash-risk measures (i.e., crash probability and return level). The derived real-time crash-risk measures are applied to estimate cycle-level crash-risk at three signalized intersections in Surrey, British Columbia. Five approaches are used to dynamically update the model parameters, including time trend model, generalized autoregressive conditional heteroskedasticity process approach, as well as the first-order, second-order, and third-order dynamic linear models. For comparison, static models are also developed. All the developed models are compared in terms of statistical fit and predictive performance. Based on the best fitted dynamic model, cycle-level crash probability and return level are calculated to measure signalized intersection safety at cycle level. The results show that dynamic models considerably outperform static models in terms of statistical fit and predictive performance. Further, the third-order dynamic model has the best performance, which is probably due to that the model incorporates two linear trends to respectively describe the variation of the coefficients as well as its change to better account for the variation in the effect of time-varying covariates. However, it should be noted that the third-order dynamic model development needs more computation time than other dynamic models, which may limit the application of the model.

This study presents a new safety measure derived from field theory. It evaluates the risk arising from the various concurrent conflicts within a platoon that can occur on high-speed highway driving situations, such as car-following, yielding, and lane changing. We defined the risk field as a finite scalar field produced by traveling vehicles on the road, and we defined the conflict field as the overlapping risk field between any vehicles in proximity on the roadway. The study used a probabilistic trajectory prediction model to construct risk fields and an approximation method to reduce the computational time for real-time applications. To demonstrate the applicability of the proposed new measure, we applied it to real-world trajectory data (NGSIM data from US Highway 101). We compared the results with three traditional conflict-based safety measures: post-encroachment time (PET), modified time-to-collision (MTTC), and deceleration rate to avoid a crash (DRAC). The new measure produced seamless and continuous risk estimations even during time windows when the other measures could not estimate the risk between vehicles. This is a major advantage over traditional measures. The study also developed visual displays of the estimated conflict fields to provide safety analysts with an intuitive and fast understanding of the results of the safety assessments made using the conflict field measure. We conclude that the proposed new safety measure provides a robust, reliable, and improved assessment of the risk involved in expected future mixed-traffic environments that involve both human-driven vehicles and automated vehicles in the future.

With the recent advancements in computer vision and artificial intelligence, traffic conflicts occurring at an intersection and associated traffic characteristics can be obtained at the granular level of a signal cycle in real-time. This capability enables the estimation of the real-time crash risk using sophisticated modelling techniques, e.g., extreme value theory. However, these models are inherently incapable of forecasting the crash risk of future time periods based on the temporal dependency of crash risks. This study proposes a unified framework of extreme value theory and autoregressive integrated moving average models for forecasting crash risks at signalised intersections. At the first level of this framework, a non-stationary generalised extreme value model has been developed to estimate the real-time rear-end crash risk at the signal cycle level using the video data collected from three signalised intersections in Queensland, Australia. To capture the time-varying effect of different traffic conditions on conflict extremes, traffic flow, speed, shockwave area, and platoon ratio covariates are incorporated into the generalised extreme value model. The signal cycle-level crash risks obtained from the first level form a univariate time series, which is modelled using two variants of autoregressive integrated moving average model to forecast the crash risk of future signal cycles. Results reveal that the autoregressive integrated moving average model with exogenous variables outperforms the model without exogenous variables and can forecast the crash risk for the next 30–35 min with reasonable accuracy. Similarly, results also demonstrate that different crash risk patterns within a typical day are accurately predicted. The proposed framework helps identify the spatiotemporal windows where safety gradually deteriorates over time, thus enabling proactive safety assessment.

Due to visibility limitations and other factors, the injuries sustained by bicyclists in nighttime vehicle-bicycle crashes tend to be more severe than daytime crashes. This paper seeks to provide insights into this day/night injury severity phenomenon by studying how day/night bicyclist injury severities have changed in crashes that occurred before, during, and after the COVID-19 lock downs. Using data from vehicle-bicycle crashes in the state of Florida over a three-year period (from 2019 to 2021 inclusive), separate yearly models of bicyclist-injury severities (with possible outcomes of severe injury, minor injury, and no visible injury) were estimated using a random parameters logit approach with possible heterogeneity in the means and variances of random parameters. Likelihood ratio tests were conducted to examine the overall stability of model estimates across the studied years as well as day/night differences, and a comparison of partially constrained and unconstrained temporal modeling approaches was undertaken. A wide range of variables potentially affecting resulting bicyclist injury severities in vehicle/bicycle crashes was considered including bicyclist and vehicle driver information, vehicle features, roadways and environmental conditions, temporal characteristics, and roadway features. The findings show statistically significant injury-severity differences between daytime and nighttime before, during and after the COVID-19 pandemic. Out-of-sample simulation results suggest that improving the visibility of bicyclist through mandated reflectivity, improved roadway illumination, undertaking public awareness campaigns relating to nighttime bicyclist safety, and vulnerable road user detection sensors in vehicles can all contribute to substantially improving nighttime bicyclist safety.

Run-off-road crashes are one of the most significant causes of road deaths worldwide. Given such significant safety concerns, a number of earlier studies examined the critical factors contributing towards run-off-road crash severity outcomes, mostly by using the information compiled in the official crash database. However, the official crash databases are less likely to have detailed information on driver behavior (errors/expectations) and roadway environment (roadway geometry and roadside attributes). This study aims to investigate the effects of design consistency measures on run-off-road crash severity mechanisms by applying a random parameters hierarchical ordered Probit model. This study contributes towards existing safety literature by demonstrating a complementary approach to capturing the effects of driver behavior and heterogeneity in roadway environment on run-off-road crash severity outcome through the composite measures of design consistency indices and cosmopolite measures of roadside hazard type variables. Specifically, 17 different functional forms of design consistency indices are developed to capture the behavioral factors from the road-geometric changes in developing run-off-road crash severity models. Further, in examining the effect of different types of the roadside environment on run-off-road crash severity outcomes, seven roadside hazard type variables are generated as a composite function of roadside object type and clear zone (lateral distance to roadside object). The empirical analysis of this study involves a two-step modelling approach - in the first step, the decision tree algorithm is applied to identify the higher-order interaction among independent variables, and in the second step, crash severity models are developed by employing several econometric approaches. The hybrid models are estimated by employing four econometric frameworks, which include Ordered Probit, Hierarchical Ordered Probit, Random Parameters Ordered Probit, and Random parameters Hierarchical Ordered Probit models. The run-off-road crash severity models are estimated by using crash data collected from the State of Queensland, Australia, for the years 2015 through 2019. Overall, this study reveals the importance of considering the interaction of drivers' behavior, road geometry, and roadside attributes along with other independent variables in developing run-off-road crash severity models.

Transportation safety studies have been mostly focused on using crash data that are rare events. Alternatively, conflict estimation can be used to assess safety. This has been proven as a proactive design methodology that does not rely on crashes and requires shorter observation. Traditionally, the safety studies involving both these reactive and proactive methods were based on aggregated data that does not take individual vehicle dynamics into consideration. This paper addresses this research gap by proposing a novel real-time conflict prediction methodology that uses previous instance trajectory data of individual vehicles to understand whether there can be potential conflict in the near future. A long-short term memory (LSTM) model is developed that can apprehend a conflict situation 9 s in the future. Data from connected vehicles have been used. The proposed model returned a recall of 81% with a false alarm rate of 28%. The predictive model has the potential to be implemented on vehicle dashboards to warn drivers of a conflict. The authors have also used SHAP (SHapley Additive exPlanation) to interpret the results from the LSTM model. It was deduced that acceleration above 0.3 m/s², deceleration within −1.5 m/s² to −0.25 m/s², and speed of more than 40kph were responsible for inducing a conflict.

The use of traffic conflict-based models to estimate crash risk and evaluate the safety of road locations is a popular direction for road safety analysis. However, a challenging issue of traffic conflict-based crash risk modeling is the selection of an appropriate sample size. Reliable conflict-based crash risk models typically require a large sample size which is always very difficult to collect. Further, when choosing a sample size, the bias-variance trade-off of model estimation is a constant concern. This study proposes an approach for identifying an adequate sample size for conflict-based crash risk estimation models. The appropriate sample size is determined by checking the model convergence and its goodness-of-fit. A quantitative approach for objectively testing the model goodness-of-fit is developed. Both the trace plots and the variation tendencies of Brooks-Gelman-Rubin statistics of parameter simulation chains are examined to inspect the model convergence. A graphical method is also used to check the model goodness of fit. If the model has not converged or fits poorly, then additional samples are required. The proposed method was applied to identify the adequate sample size for a Bayesian hierarchical extreme value theory (EVT) block maxima (BM) model using traffic conflict data from four signalized intersections in the city of Surrey, British Columbia. The indicator, modified time to collision (MTTC), was used to delineate traffic conflicts. A series of stationary and non-stationary Bayesian hierarchical BM models were developed using the cycle-level maximums of negated MTTC. The adequate sample sizes of stationary and non-stationary Bayesian hierarchical BM models were determined separately. Further, two methods of increasing the sample size (i.e., extending the observation period and combining data from different sites) were compared in terms of goodness-of-fit as well as crash estimate accuracy and precision. The results show that for both stationary and non-stationary models, the sample size used is adequate for model convergence and goodness-of-fit. Moreover, adding covariates to the stationary Bayesian hierarchical BM model does not affect the size of the required sample. Extending the observation period outperforms combining data from different sites in terms of goodness-of-fit as well as crash estimation accuracy and precision of non-stationary models. This is likely related to the existence of unmeasured factors that could impair model estimation and inference when merging data from several sites to augment the number of samples. Overall, the findings of this study can be applied to examine whether available data is adequate and the amount of additional data required for producing reliable statistical inference.

Several studies have shown that crash risk is a dynamic quantity that is frequently changing with considerable spatial and temporal variations. Recent advances in safety evaluation techniques such as using extreme value theory (EVT) models provided the opportunity to use traffic conflict data obtained from road user trajectories to estimate real time safety metrics. These metrics can aggregate crash risk along a certain route based on the duration of exposure to unsafe road conditions. This paper applies a Bayesian hierarchal extreme value theory model to trajectories obtained from a drone dataset from Athens, Greece, to develop a safest route algorithm capable of informing users about the safest route in an urban network in real time. The study area selected consists of a rectangular grid made up of 102 signalized and unsignalized intersections. The dynamic crash risk for each link in the network was obtained and used to identify the safest route between any origin–destination pair and the corresponding fastest route. The safest routes were then compared to the fastest routes and were found to be 22% safer on average, resulting in an 11% increased travel time. Moreover, the safest route was identical to the fastest route in 23% of the origin–destination pairs analyzed and had an average similarity of 54% in terms of links. Recognizing the trade-off between safety and mobility, a multi-objective routing methodology was proposed which balances travel time and crash risk using a weighted preference for safety. This work has considerable potential for improving the safety of all road users and may also be used for fleet routing applications as part of multi-objective routing systems.