Accident; analysis and prevention最新文献

Driving behavior is crucial in shaping traffic dynamics and serves as the foundation for safe and efficient autonomous driving. Despite the widespread interest in driving behavior modeling, existing models often focus on specific behaviors and cannot describe all types of vehicle movements, while vehicle status and driving scenarios are dynamic and infinite. That means comprehending and modeling generalized driving behavior mechanisms is essential. Risk Homeostasis Theory (RHT) emerges as a compelling conceptual framework to explain human risk behaviors comprehensively. The critical problem in modeling behavior using RHT is quantifying the subject risk precepted by humans. RHT has been applied in car-following behavior modeling based on the one-dimensional risk indicator Safety Margin (SM), simplifying the specific behavior along its direction. While the generalized perceived risk indicator on the two-dimensional surface still lacks. Considering the collision avoidance capacity from the driver's perspective, this paper proposes the two-dimensional safety margin (TSM) to describe the driver's risk perception in generalized driving scenarios with two-dimensional movements. Results demonstrate that TSM could accurately describe car-following behavior compared to existing risk indicators, with a 9.1 % correlation improvement and the reasonably calibrated response time (1.07 s). And TSM could effectively capture the discrepant risk perceptions of different drivers involved in the same conflict, underscoring the alignment of TSM with drivers' subjective risk perceptions. Besides, TSM reflects the risk homeostasis of driving behaviors, as both typical scenarios have the normally distributed and concentrated target levels. Further, TSM also achieves a generalized, scenario-independent risk quantification with a mean target level of 0.85. As a good representation of driver's risk perception in two-dimensional scenarios, TSM serves as a crucial basis in areas such as driving behavior modeling, and decision-making and testing of autonomous driving.

Cyclists are among the most vulnerable road users, increasingly subject to various sources of distraction, including the use of mobile phones and engagement in other tasks while navigating urban environments. Understanding and mitigating the impact of these distractions on cyclist safety is crucial. Despite the importance of this issue, the effect of distraction on injury severity in cycling crashes has not been extensively studied. This research analyzes four years of U.S. crash data (2019-2022) from the Crash Report Sampling System (CRSS) database, employing a hybrid framework that integrates CatBoost-based SHAP algorithm and the random parameters binary logit model with heterogeneity in means and variances (RPBL-HMV). The proposed approach confirms the significant role of cyclist distraction in crash injury severity. Subsequently, the analysis identifies several factors influencing the likelihood of severe injuries in distracted cyclist crashes. Crashes involving the front of motor vehicles, occurring in rural areas, on two-way roads, at higher speed limits, and during weekends were associated with a higher probability of severe injuries. Conversely, crashes at T-intersections, involving the side or rear of motor vehicles, where cyclists wore helmets, or during rush hour were linked to a reduced likelihood of severe injuries. Notably, interaction effects reveal nuanced patterns. For instance, while crossing roadway actions and rush hour periods individually decrease the likelihood of severe crashes, their combination increases the probability of such outcomes. The findings suggest targeted safety measures and policy interventions aimed at enhancing cyclist safety and promoting safer cycling environments by mitigating distraction-related risks.

Shared spaces prioritise the role of micromobility in urban environments by separating vulnerable road users from motorised vehicles, aiming to enhance both actual and perceived safety. However, the presence of various transport modes, such as pedestrians, cyclists and e-scooters, with differing navigation behaviours, increases the heterogeneity of these spaces and impacts the perception of safety. Despite the increasing use of e-scooters, the safety perceptions of e-scooter riders remain largely underexplored in the literature. In response, I conducted an online video experiment and polled 920 e-scooter users in Sweden to assess their safety perceptions when interacting exclusively with cyclists. I collected data on socio-demographics, travel habits, crash history, and responses to hypothetical video scenarios depicting interactions in shared spaces, where e-scooter riders overtake or meet cyclists. I then employed a random effect latent class ordered logit model to quantify the determinants of e-scooter riders' safety perceptions. The findings indicate that women feel less safe in shared spaces compared to men. Additionally, the direction of encounters significantly affected young adults, who perceived meeting other users as more unsafe than overtaking them. These findings highlight the importance of accounting for unobserved heterogeneity in safety perceptions, emphasise the significant role of demographic variables in understanding users' safety perceptions, and reinforce the need for inclusive design of shared spaces for all road users.

Intersections are frequently identified as crash hotspots for roadways in major cities, leading to significant human casualties. We propose crash likelihood prediction as an effective strategy to proactively prevent intersection crashes. So far, no reliable models have been developed for intersections that effectively account for the variation in crash types and the cyclical nature of Signal Phasing and Timing (SPaT) and traffic flow. Moreover, the limited research available has primarily relied on sampling techniques to address data imbalance, without exploring alternative solutions. We develop an anomaly detection framework by integrating Generative Adversarial Networks (GANs) and Transformers to predict the likelihood of cycle-level crashes at intersections. The model is built using high-resolution event data extracted from Automated Traffic Signal Performance Measures (ATSPM), including SPaT and traffic flow insights from 11 intersections in Seminole County, Florida. Our framework demonstrates a sensitivity of 76% in predicting crash events using highly imbalanced crash data along with real-world SPaT and traffic data, highlighting its potential for deployment at smart intersections. Overall, the results provide a roadmap for city-wide implementation at smart intersections, with the potential for multiple real-time solutions for impending crashes. These include adjustments in signal timing, driver warnings using various means, and more efficient emergency response, all with major implications for creating more livable and safe cities.

Mixed platoon with a human-driven leading vehicle may be a transition mode prior to the widespread adoption of fully autonomous platoon. Enhancing the driving safety of the leading vehicle driver is crucial for improving the overall operational safety of the mixed platoon. Predictive-Forward-Collision-Warning (PFCW), an emerging technology in transportation, holds promise in mitigating collision risks for drivers by presenting traffic information beyond their immediate visual range. However, the influence characteristics of this function and how it influences the evolution of collision risk in leading vehicle driver remain unclear. Therefore, this paper attempts to analyse the quantitative impact of PFCW on the collision risk of leading vehicle driver. A test platform for connected mixed platoon was built utilizing driving simulation technology, alongside the development of a connected Human-Machine Interface (HMI) incorporating PFCW functionality. To evaluate the longitudinal collision risk of leading vehicle driver, a time-frequency analysis method was employed, focusing on key indicators: deceleration rate to avoid collision (DRAC), time to collision (TTC), and proportion of stopping distance (PSD). The time-domain analysis results indicated that PFCW can significantly mitigate the collision risk of leading vehicle. Wavelet transform results demonstrated that PFCW can ameliorate drivers' abnormal driving behavior and mitigate the collision risk in emergency situation of impending collision moment. Meanwhile, PFCW can enhance the overall operation safety of the mixed platoon. This paper leverages driving simulation technology and multidimensional indicators to analyze the quantitative impact of PFCW on the collision risk of leading vehicle driver during rapid deceleration of preceding vehicles. The findings can guide the development of test standards for connected mixed platoon, the promotion and application of PFCW, and the advancement of Navigate on Autopilot (NOA). Additionally, the test platform and framework developed in this study can accommodate various experimental needs for connected mixed platoon testing.

Intelligent safety systems (ISS) for autonomous vehicles, integrating advanced perception capabilities and passive protection devices, are expected to reshape traditional pedestrian safety systems and play a key role in reducing the risk of pedestrian injuries in traffic accidents. However, traditional active control and passive protection modules remain disconnected due to insufficient evidence supporting the effectiveness of collaborative strategies in integrated systems, particularly concerning activation criteria and timing. This study aims to address this gap by developing a comprehensive ISS that incorporates advanced perception systems, a vehicle dynamic control module, and controllable passive safety devices. Furthermore, the study evaluates the efficacy of trigger strategies in minimizing injury risks in various safety systems including Automatic Emergency Braking (AEB), Automatic Emergency Steering (AES), and ISS. To achieve this, we reconstructed the dynamics of pedestrian-vehicle interactions before collisions by examining 23 detailed collision cases. These cases were selected from real-world accident databases and included clear video recordings and detailed injury reports. Additionally, we analyzed the boundary conditions for collision avoidance by constructing vehicle steering and braking avoidance models. Our findings indicate that, in real-world accidents, the average Time-to-Collision (TTC) required for drivers to avoid collisions is -3.15 ± 1.00 s. In contrast, the AEB system requires -1.06 ± 0.23 s, and the AES system requires -0.44 ± 0.14 s. Building on this, we developed injury risk models for the system activation, predicting collision risks at various TTCs and pedestrian injury risks. The pedestrian injury risk prediction model effectively forecasts the risk of AIS3 + head injuries resulting from collisions between pedestrians aged 20 to 70 years and the vehicle hood. The threshold for a severe AIS3 + head injury risk is set at 10 %, with a trigger TTC of the ISS at -0.60 ± 0.20 s. When the system is activated at a TTC of -0.5 s, it can reduce the probability of severe head injury to pedestrians by 59 %. The design of the ISS shows significant potential for enhancing pedestrian safety. The findings of this research can offer guidance for the activation strategies of passive safety devices based on input signals from advanced perception systems in AVs.

Speeding crashes remain high injury severities after the stay-at-home order in California, highlighting a need for further investigation into the fundamental cause of this increment. To systematically explore the temporal impacts of the stay-at-home order on speeding behaviors and the corresponding crash-injury outcomes, this study utilizes California-reported single-vehicle speeding crashes on freeways (access-controlled) and non-freeways (non-access-controlled) before, during, and after the order. Significant injury factors and in-depth heterogeneity across observations are identified by random parameter logit models with heterogeneity in means and variances. Without segmenting the data by periods, the partially temporally constrained approach is employed to statistically determine varying and stabilized parameters over time through the whole dataset. Different likelihood ratio tests reveal significant temporal instabilities and stabilities of factors between two roadways and three periods. The potential impacts of observation selection issues on the marginal effect calculations of the partially constrained models are also systematically investigated. Significant variations in the probability of severe injury rate per week after the order are also found based on the Mann-Whitney U tests. The hysteretic effects of the order on the crash frequency and severity are observed on both freeways and non-freeways. Overall, seven variables are found to have stable effects, while fifteen variables exhibit unstable effects over time. Significant temporal variations in driver behaviors, including driving under the influence, cell phone usage, hit-and-run, failure to use seat belt, entering or leaving the ramp, and reaction to previous collisions, are observed before, during, or after the order. Specific countermeasures and effects of heterogeneity in means and variances are also discussed. These findings provide insights into understanding the temporal impacts of the stay-at-home order on injury severities, which are valuable to decision-makers and researchers for future order practice, restriction improvement, and complementary policy development.

Cooperative control of intersection signals and connected automated vehicles (CAVs) possess the potential for safety enhancement and congestion alleviation, facilitating the integration of CAVs into urban intelligent transportation systems. This research proposes an innovative deep reinforcement learning-based (DRL) cooperative control framework, including signal and speed modules, to dynamically adapt signal timing and CAV velocities for traffic safety and efficiency optimization. Among the DRL-based signal modules, a traffic state prediction model is merged with the current state to augment characteristics and the agent-learning process. A multi-objective reward function is designed to evaluate safety and efficiency using a traffic conflict prediction model and vehicle waiting time. The double deep Q network (DDQN) model is used to design the agent observing the traffic state, learning the optimal signal control policy, and then inputting the signal phase into the speed module. Based on the green duration analysis and constraints of mixed traffic flow of CAVs and human-driven vehicles, a speed planning model is constructed to optimize CAVs' speed and alter traffic state, which in turn affects the agent's next signal decisions. The proposed framework is tested at isolated intersections simulated by two real-world intersections in Changsha, China. The results reveal the superiority of the proposed method over DRL-based traffic signal control (DRL-TSC) in terms of coverage speed and computation time. Compared to actuated signal control, adaptive traffic signal control, and DRL-TSC, the proposed method significantly optimizes traffic safety and efficiency across diverse intersections, temporal spans, and traffic demands. Furthermore, the advantage of the proposed method substantially amplifies with the increased CAV penetration, regardless of the intersection types.

Near-miss traffic risk estimation using Extreme Value Theory (EVT) models within a real-time framework offers a promising alternative to traditional historical crash-based methods. However, current approaches often lack comprehensive analysis that integrates diverse roadway geometries, crash patterns, and two-dimensional (2D) vehicle dynamics, limiting both their accuracy and generalizability. This study addresses these gaps by employing a high-fidelity, 2D time-to-collision (TTC) near-miss indicator derived from autonomous vehicle (AV) sensor data. The proposed framework uses univariate Generalized Extreme Value (UGEV) distribution models applied to a subset of the Waymo motion dataset across six arterial networks in San Francisco, Phoenix, and Los Angeles. Extreme events are identified through the Block Maxima (BM) sampling-based approach from each conflicting pair, with 20s block sizes to account for the scarcity of samples in short-duration traffic segments. The framework also incorporates conflicting vehicle dynamics (e.g., speed, acceleration, and deceleration) as covariates within a non-stationary hierarchical Bayesian structure with random parameters (HBSRP) UGEV models, allowing for the effective management of vehicle spatial, temporal, and behavioral heterogeneity. Results show that HBSRP-UGEV models outperform other approaches, with a 6.43-10.56% decrease in DIC, especially for near-miss events in short-duration traffic segments. The inclusion of dynamic vehicle behaviors and random effects substantially enhances the model's capability to estimate real-time traffic risks. This generalized real-time EVT model bridges the gap between active and passive safety measures, offering a precise and adaptable tool for network-level traffic safety analysis.