Minhua Shao

The increasing deployment of connected and automated vehicles (CAVs) challenges the behavioral‐homogeneity assumption of classical traffic‐flow theory, yet direct empirical evidence on how microscopic longitudinal driving behavior varies across vehicle‐interaction modes in real‐world mixed traffic remains limited. Existing simulation and driving‐simulator studies rely on idealized CAV control assumptions, while trajectory‐based studies frequently adopt inverse calibration of predefined car‐following models, introducing model‐structure assumptions, parameter non‐identifiability, and equifinality. Using same‐source unmanned aerial vehicle (UAV) trajectory data from an urban open‐road test segment, this study develops a regime‐based empirical framework to identify and quantify four driving‐behavior parameters: minimum time headway ( T _min ) in the steady car‐following regime, minimum standstill net gap ( s _min ) in the full‐stop regime, maximum deceleration ( d _max ) in the closing‐in car‐following regime, and maximum acceleration ( a _max ) in the free‐flow acceleration regime. For each parameter, regime‐specific trajectory episodes are identified, values are extracted using within‐regime quantile‐based statistics, and threshold sensitivity is described by comparing sample retention and parameter medians across 196 discrete candidate combinations. Empirical distributions are profiled in terms of central location, dispersion, skewness, kurtosis, and tail behavior; cross‐mode differences are quantified using the first‐order Wasserstein distance and Cliff’s δ . In the observed mixed‐traffic setting, behavioral heterogeneity manifests not only through mean‐level shifts but also through systematic distributional reshaping. Relative to the HDV–HDV reference group, CAV–HDV interactions exhibit a substantial upward shift and stronger concentration in T _min , whereas HDV–CAV interactions show larger and more dispersed s _min . For d _max , both heterogeneous interaction modes attenuate the high‐intensity braking tail, with HDV–CAV showing a more pronounced downward shift. In the free‐flow acceleration regime, CAVs exhibit slightly higher and more concentrated a _max values than HDVs. These findings provide same‐source empirical evidence for microscopic longitudinal driving‐behavior heterogeneity and offer parameter‐level references for interaction‐mode‐specific simulation, tail modeling, capacity analysis, and disturbance‐propagation studies.

Traffic flow forecasting is a strongly supportive component of intelligent transportation services. While in light of the expanding road networks or city grids, there is a critical concern to enhance both the accuracy and efficiency of prediction models. Despite the remarkable improvements in prediction accuracy, existing research continues to face three limitations in practical engineering scenarios. Firstly, current research often overlooks the time delay characteristics when capturing spatial relationships between global nodes. Secondly, most approaches have a quadratic computational complexity with respect to the number of nodes, resulting in significant training overhead and poor scalability. Furthermore, studies that do consider dynamic spatial relationships typically require complex model structures, resulting in higher computational costs. To address these issues, we propose a Time-Lag Aware Spatial-temporal Transformer (TLAST), a lightweight yet effective traffic flow forecasting model. TLAST introduces a cross-time strategy into the embedding stage and the attention extraction to capture the time-lag aware spatial-temporal features. Furthermore, we propose a Spatial Proxy Attention (SPA) module. It utilizes proxy representations to efficiently capture time-varying spatial dependencies with linear complexity, significantly reducing computational overhead. Extensive experiments on seven real-world traffic datasets demonstrate that TLAST consistently outperforms state-of-the-art baselines, achieving up to 7.84% improvement in prediction accuracy (MAE) while reducing memory usage and time cost by 85.21% and 75.14%, respectively. Results from the empirical analysis not only demonstrate the model’s efficiency and scalability but also highlight its practical usability in real-world traffic forecasting scenarios.

Estimating average travel speed (ATS) on urban streets is essential for travel planning, traffic guidance, and congestion control. Floating car data has been widely applied in estimating ATS on urban streets. The harmonic average of the floating car speed is mostly used as the true average speed in research and practice. However, the estimation accuracy of this practice is affected by the travel time distribution (TTD) and the sampling rate of floating cars, especially when facing the interrupted flow on urban streets. To address this problem, this work develops a framework for estimating ATS of urban streets based on TTD. We analyze the characteristics of TTD by using the fully sampled trajectory data observed on urban roads in Shanghai, China. It is found that the travel time of stopped and unstopped vehicles show different statistical distribution characteristics, which should be considered in estimating ATS. The focus of this method is to divide vehicles on urban roads into stopped vehicles and unstopped vehicles, and to study the distribution laws of their travel times. Based on these laws, we improve the existing ATS estimation method. In addition, this paper further discusses the sampling rate of floating cars for estimating ATS by the random sampling method. The results show that the recommended sampling rate is 5%–7%, which is slightly higher than the sampling rate of 3%–5% in the existing research. Finally, the experiment based on empirical data substantiates the superior performance of the proposed method in comparison to the harmonic average method.

Deep Reinforcement Learning (DRL) is capable of learning a policy with great scene adaptation ability through interactions with the environment, and has application potential in the field of autonomou...

Due to the escalating transportation demand and the significant ramifications of traffic congestion, there is an imperative to investigate the sources of congestion, known as “congestion bottlenecks”. The implementation of control methods ahead of the occurrence of congestion is crucial. Connected and autonomous vehicles (CAVs) have a high potential within the field of traffic control. CAVs are exceptionally controllable and facilitate management feasibility. This study utilizes the high compliance of CAVs to provide an effective solution for the congestion management problem at the network level when mixed traffic flows are saturated. A linear programming model to reduce average travel time over the road network is developed. By utilizing a genetic algorithm, the optimal traffic demand regulation scheme can be obtained and the departure time of CAVs optimized. The effectiveness of the proposed method is validated through simulation across various road network scales, CAVs penetration rates, and controlled CAV proportions. The proposed method can only control a specific amount of CAVs, which, according to an analysis of the simulation results, significantly improves the performance of the transportation system. The importance of employing advanced control methods to improve the sustainability of urban transportation development and the travel experience is underscored in the conclusion.

The research on price elasticity of demand, especially in the field of transportation, has high theoretical and application value. Based on the perspective of price elasticity of demand, the study presents the impact of adjusting expressway rates on the traffic flow of cars with seven seats or less. The data are from the measured data of the Shanghai expressway Electronic Toll Collection (ETC) from 2019 to 2020. In order to eliminate the impact of the surge of ETC users in 2019 on the results, Empirical Mode Decomposition (EMD) is used to optimize the data. The research shows that the price elasticity of demand will increase with the increase in charge amount (distance).

With the increase in the demand for and transportation of hazardous materials (Hazmat), frequent Hazmat road transport accidents, high death tolls and property damage have caused widespread societal concern. Therefore, it is necessary to carry out risk factor analysis of Hazmat transportation; predict the severity of accidents; and develop targeted, extensive and refined preventive measures to guarantee the safety of Hazmat road transportation. Based on the philosophy of graded risk management, this study used a priori algorithms in association rule mining (ARM) technology to analyze Hazmat transport accidents, using road types as classification criteria to find rules that had strong associations with property-damage-only (PDO) accidents and casualty (CAS) accidents under different road types. The results indicated that accidents involving PDO had a strong association with weather (WEA), traffic signals (TS), surface conditions (SC), fatigue (FAT) and vehicle safety status (VSS), and that accidents involving CAS had a strong association with VSS, equipment safety status (ESS), time of day (TOD) and WEA when urban roads were used for Hazmat transportation. Among Hazmat transport incidents on rural roads, the incidence of PDO accidents was associated with intersections (IN), SC, WEA, vehicle type (VT), and segment type (ST), while the occurrence of CAS accidents was associated with qualification (QUA), ESS, TS, VSS, SC, WEA, TOD, and month (MON). Strong associations between the occurrence of PDO accidents and related items, such as IN, SC, WEA and FAT, and the occurrence of CAS accidents and related items, such as ESS, TOD, VSS, WEA and SC, were identified for Hazmat road transport accidents on highways. The accident characteristics exemplified by strongly correlated rules were used as the input to the prediction model. Considering the scarcity of these events, four prediction models were selected to predict the severity of Hazmat accidents on each road type employing four analyses, and the most suitable prediction model was determined based on the evaluation criteria. The results showed that extreme gradient boosting (XGBoost) is preferable for predicting the severity of Hazmat accidents occurring on urban roads and highways, while nearest neighbor classification (NNC) is more suitable for predicting the severity of Hazmat accidents occurring on rural roads.