Journal of Transportation Engineering, Part A: Systems

Max-pressure control is a new approach to signal timing with mathematically proven network throughput properties. Over the past decade, max-pressure control has emerged from a novel mathematical idea in a simple store-and-forward queueing model to include practical aspects like a signal cycle, realistic traffic flow models, measurement errors, and intersection access for alternative modes. Furthermore, max-pressure control is adaptive to local queues and to changes in network demand. Max-pressure control is also decentralized and easy to compute. A broad range of numerical results in calibrated microsimulation models have mostly demonstrated superior performance from max-pressure control compared with existing signal timing methods, and some researchers have started to experiment with max-pressure control on actual roads. Given these benefits, the purpose of this review paper is to provide a summary of the mathematical approach, methodological improvements, and numerical results. This summary is intended for researchers interested in continuing methodological or numerical work and for practitioners exploring the potential use of this state-of-the-art signal timing method.

This article proposes a methodology to compute the annual parking permit fees for university employees on campuses. This methodology determines separate permit fees for faculty and staff for each parking zone. It first establishes the campus-wide base prices for employees, faculty, and staff parking permits. A tobit regression model, using the data from 213 universities across the United States, was developed to estimate the employee base price of a campus. The employee base price is then combined with the median salaries of faculty and staff, respectively, to derive the faculty and staff base prices. The methodology then assumes that the employees, faculty, and staff each have a constant value of parking access. For each zone, the annual value of parking access is the sum of the annual permit fee of the zone and the annual value of walking time from the zone to the subject’s workplace on campus. Knowing the average annual value of parking access and the average annual value of walking time from a zone, the average annual parking permit fee of the zone can be determined. The methodology is applied to a campus that has 41 employee parking zones in an example problem.

Abstract

In this paper, a framework of active obstacle avoidance for autonomous vehicles based on the hybrid soft actor-critic (SAC) algorithm is proposed. In the stage of local path planning, a comprehensive cost function considering collision risks, deviation from the global route, road lines crossing, and driving comfortability is developed to provide a local optimal path avoiding both static and dynamic obstacles considering multiple predicting timesteps. Then, a path tracking controller on the foundation of a hybrid SAC algorithm is designed to mitigate the problem of high sample complexity caused by random initialization of parameters in conventional reinforcement learning approaches. Model predictive control (MPC) plays a guiding role by applying its control action to combine with the action of SAC online to obtain a more effective state and reward information for training. The mechanism of the combination of MPC with SAC to balance the exploration and reliability is explained in detail. In order to improve the convergence rate and learning efficiency, a dual actor network structure for two different control actions is adopted. With considerations of various relevant factors influencing the control effect, the reward for the hybrid SAC algorithm is designed carefully. Finally, the results of simulation experiments illustrate that the proposed approach performs effectively with the assurance of safety and driving comfortability. In summary, the hybrid SAC algorithm with dual actor networks performs better than other algorithms for comparison in all test scenarios in this paper.

Practical Applications

An autonomous vehicle is an attractive topic with the development of artificial intelligence. Obstacle avoidance is a significant challenge to autonomous vehicles, which is also the focus of this paper. To promote the reliability of autonomous vehicles in the real driving environment, this paper proposes a framework of active obstacle avoidance for autonomous vehicles based on the hybrid soft actor-critic algorithm divided into two steps of local path planning and path tracking. In the first step, a comprehensive cost function considering various factors is developed to provide a local optimal path avoiding both static and dynamic obstacles, which can be embedded into the planning module of autonomous vehicles and achieve real-time collision-free path planning by an onboard computer. In the second step, the proposed hybrid soft actor-critic algorithm for path tracking control can solve the problem of low sample efficiency of conventional deep reinforcement learning. In addition, it is trained in the simulation environment without the risk of collisions in a real driving environment and could be migrated to the real vehicle platform conveniently as it is designed based on the vehicle dynamics model. The proposed approach is both academic and practical and can promote the development of autonomous driving in the real world.

Abstract

The transportation system undergoes severe impacts due to potholes and the presence of stray animals on the roads resulting in accidents and fatal injuries. The utilization of intelligent transportation systems would reduce accidents and impart safety to the overall transportation network. This research aims to impart transportation safety through a real-time alert warning system for avoiding accidents due to potholes and the presence of stray animals. The study incorporates real-time detection of transportation entities like vehicles, animals, and pedestrians through a YOLO v3 computer vision algorithm processed on the GPU environment for a higher frame rate. The potholes and animal hotspots are mapped to form a geospatial database on which the buffer tool of geographic information system (GIS) is applied. The buffer zone was implemented on the geospatial layer to alert the driver in real-time, while the vehicle approaches the buffer zone. The system yields high precision of 0.976 mean average precision (mAP) score of entity detection and the real-time alert warning alerts the driver to ensure transportation safety while avoiding any possible accidents or fatal crashes.

Practical Applications

Increasing population density gives rise to traffic congestion, haphazard movement of vehicles, and accident-prone scenarios. This results in the loss of countless lives, which is undesirable. Animals and potholes are the major contributors to accidents, and minimizing them is necessary for maintaining smooth and safe transportation. Advanced driver assistance systems (ADAS) play a crucial role in reducing accidents and increasing safety in the transportation network. This paper presents an integrated system of enhanced ADAS functionalities like the detection of animals and alerting the driver about potential animal hotpots and pothole areas in real-time. The system utilizes deep learning techniques for real-time object detection and GIS-based techniques for warning the drivers of potholes and animal zones in advance. This alert the drivers making them in control of the vehicle for avoiding any probable crash. The system induces safety in the transportation network and opens a strong scope for self-driving cars and connected vehicle infrastructure.

Highway congestion is a common problem worldwide, and different management strategies have been proposed and implemented. In Taiwan, commuter trips in urban areas tend to use freeways for daily trips. One particular problem in Taiwan is consecutive on-ramps within metropolitan areas, such as Taipei, Taichung, and Kaohsiung. Three to four on-ramps bring more vehicles to the freeways, and coordinating with an integrated ramp control strategy is critical to avoiding congestion in metropolitan areas. Based on the concept of the store-and-forward method, this study proposes a linear optimal control model to optimize integrated on-ramp control strategies. The store-and-forward method is chosen because freeway congestion is usually observed in these areas, and freeway efficiency depends on avoiding queue length. The developed mathematical program can be solved efficiently through mathematical optimization packages. The objective of the formulation is to minimize the total queue length within considered areas. In the Taipei metropolitan area, numerical experiments were conducted based on three consecutive on-ramps, including Yuanshan, Taipei, and Sanchun ramps. Several scenarios were designed to illustrate the possible benefits of the proposed model.

The on-ramp area has been a hot spot for research as a bottleneck area. In this paper, an on-ramp cooperative merging control strategy is proposed to make full use of active cooperative vehicles, and alleviate the negative influence of merging on the main road. First, a two-stage framework is proposed to illustrate the merging strategy. At the vehicle level, the cooperative merging controllers of vehicles are designed based on an optimization algorithm that considers the penalty function. At the traffic level, the cooperative distance of the upstream and downstream vehicles at the main road target merging position are optimized to improve merging efficiency. Meanwhile, to ensure the implementation of the proposed strategy in mixed traffic, we propose a method to determine the lead vehicles based on the proximity principle. Numerical experiments were performed to verify the fully proactive and cooperative merging (FPCM) strategy, the effectiveness of the vehicle cooperative controllers, the necessity of optimizing the cooperative distance, the effect of different velocities on the main road and ramp, and the effectiveness of determining the lead vehicles. In the case of multivehicle merging, the proposed strategy can improve merging efficiency by 44.42% and save fuel consumption by 39.11%. In the case of a single merging vehicle, the optimized proposed strategy can improve merging efficiency by 48.4% and save fuel consumption by 31.4%. Finally, the simulation experiments for the different maximum speeds of vehicles show that a sufficient speed adjustment range can improve the advantages of the strategy.

In this work, a conserved higher-order (CHO) macroscopic traffic flow model is extended to road networks. We first introduce the Riemann problem at a junction for the CHO model and then provide a general framework to solve it, considering the compatibility between the CHO and Lighthill-Whitham-Richards (LWR) models. Through the presented framework, Riemann solvers for the LWR model can be extended to the CHO model. Specifically, a kind of Riemann solver for the LWR model is extended to derive the Riemann solver for the CHO model, in which the total actual flow at the junction is maximized. The Riemann solvers for three typical junctions (one-in-two-out junction, two-in-one-out junction, and two-in-two-out junction) are discussed. The first-order finite-volume method is adopted to solve the extended model. Numerical examples are given to validate the extended model and the numerical algorithm.

To explore traveler transfer decisions for different purposes, a transfer decision model allowing psychological comprehensive perception value is built based on cumulative prospect theory. Combined with the value function of the arrival time and time value model, the cost function of psychological comprehensive perception is established, and the reference point of psychological comprehensive perception is set. Travel cumulative prospect models of bus interchanges and bus transfers to subways chosen by commuters and noncommuters are established, and the two-dimensional (departure time and travel mode) optimal travel decisions of commuters and noncommuters are obtained based on the calculation results. The results show that traveler cumulative prospect value first increases and then decreases with the delay of departure time, and the peak value’s occurrence time of a bus transfer to a subway is later than that of a bus interchange. Cumulative prospect value decreases as the transfer time increases when travelers’ arrive at the destination at the same time. Commuters obtain higher gains when they choose late departure time and bus transfer to the subway with determined transfer time, while noncommuters obtain higher gains with the opposite choice. The results show that traveler comprehensive psychological perception not only depends on arrival time but also depends on departure time, travel time in different stages, and cost. Travelers have different risk preferences for different travel purposes, and commuters’ time value is high, which determines whether they tend to pursue risk. Noncommuters tend to avoid risk. This conclusion can provide a theoretical basis for transfer decisions to improve satisfaction with public transit.

Automated platooning is gaining increasing attention because it is regarded as a highly efficient transportation mode to solve traffic congestion problem. Although different intraplatoon spacing strategies have been proposed, little research has focused on quantitative optimization of spacing policy from a safety perspective considering differences between automated and human-driven platooning. In addition, there still is a lack of comprehensive research into string stability and delay robustness of cooperative adaptive cruise control (CACC) for heterogeneous platoons. This paper proposes a quantitative spacing policy in quadratic form based on safety index optimization under hard braking scenarios. The expected value of collision severity is taken as the main index for spacing optimization. The CACC strategy then tracks the proposed spacing considering fluctuation of multitime delays and uncertainty of vehicle dynamics. Based on sliding model control theory and frequency stability criteria, the sufficient conditions satisfying CACC stability and string stability were given. The robustness of the CACC system considering communication latency fluctuation was shown using simulation results. Compared with the existing control strategy, the proposed control strategy is more robust to communication latency, and string stability can be guaranteed. The proposed spacing strategy effectively can reduce collision risk, and the proposed CACC strategy improves latency robustness on the premise of ensuring string stability.

When traffic congestion occurs on freeway off-ramp bottlenecks, the traffic state becomes complicated and changeable, which leads to increased vehicle travel time and decreased traffic safety and traffic efficiency. Variable speed limit (VSL) control is an effective method to improve traffic conditions, increase bottleneck throughput, improve traffic efficiency, and reduce emissions. Currently, there is an emerging trend of using connected and autonomous vehicle (CAV) technology to develop VSL control. This paper proposes an optimal differential variable speed limit (DVSL) control strategy under mixed CAVs and human-driven vehicles (HVs) environment for freeway off-ramp bottlenecks. The proposed DVSL control considers the characteristics of on-ramp, off-ramp, and mixed traffic flow (i.e., CAVs coexist with HVs). The proposed optimal DVSL control can describe and forecast the dynamics of traffic flow, and can set different speed limits across each lane with a multiple-objective function of total travel time (TTT) and total travel distance (TTD). A model predictive control (MPC) approach was utilized to optimize the DVSL control algorithm. The designed DVSL control was tested on a real-word freeway section with a simulated off-ramp bottleneck. The simulation results show that the proposed control strategy outperforms other existing methods in terms of improving the mobility of a freeway off-ramp bottleneck and maximizing the environmental benefits. Sensitivity analysis shows that the proposed control strategy can improve performance with the increase of the penetration rate (PR) of CAVs. The proposed methods form the basis of VSL control at off-ramp sections under mixed traffic environment.

Car-following models are calibrated to account for various driver behaviors such as speed and space headway. Because drivers do not all drive the same way, they are typically classified based on their level, or profile, of aggressiveness. This approach to model calibration assumes that a single set of car-following parameters applies to an individual driver consistently, that is, during the entire driving period a study considers. The purpose of this research is to challenge that assumption by analyzing the heterogeneity and inconsistency of driver behaviors in different traffic conditions. A total of 3,262 urban expressway car-following periods from 51 drivers were extracted from the Shanghai Naturalistic Driving Study database. The intelligent driver model (IDM) was selected as this study’s car-following model for its favorable performance in highly dynamic situations; its parameters, that is, desired speed, desired time headway, maximum acceleration, comfortable deceleration, acceleration exponent, and minimum spacing, were calibrated. To reflect the effect of traffic conditions, the car-following periods were categorized into three regimes from low-speed to high-speed. Two-way analysis of variance was used to evaluate heterogeneity and inconsistency in the car-following parameters. The findings of this study show that (1) the main significant variable that distinguished driver profile was desired time headway, while the comfortable acceleration was commonly affected by both driver profile and speed regime; (2) variability of individual driver parameter estimates greatly depended on the traffic regime (low-, medium-, and high-speed), showing that driver behavior was highly influenced by traffic conditions; and (3) speed information was demonstrated to be a useful tool by which to divide car-following periods for modeling, as the car-following parameters exhibited distinct differences among the regimes in the varied traffic environment.

Car-following models serve an important role in gaining a thorough understanding of traffic flow and driving behavior characteristics. By analyzing these characteristics, the models are critical to microscopic traffc simulation, and consequently, to traffic safety. However, lack of reliable traffic data in China has, until recently, limited the use of car-following models. As the Shanghai Naturalistic Driving Study (SH-NDS) has now made such data accessible, car-following models have been built for freeways and urban expressways, but none have yet been developed for urban streets. To compare car following for the three road types and to determine the best model for urban streets, five commonly used car-following models were calibrated and validated with 5,500 urban street-level car-following events extracted from the 161,055 km of data collected in the SH-NDS. The models were evaluated based on their parameter estimates and root mean square percentage errors (RMSPE). Results show that (1) the intelligent driver model (IDM), with a calibration error of 24% and a validation error of 28%, performed best in modeling drivers’ car-following behavior on urban Shanghai streets; and (2) in comparison to previous car-following research on Chinese freeways and urban expressways, drivers on urban streets tend to assume a relatively lower car-following speed, and maintain slightly larger time headway and maximum acceleration. Because the IDM demonstrated great performance on expressways, freeways, and urban streets in China, it is reasonable to assume the model may show similar performance when used to analyze car following in other countries.

Accurate prediction of the change in track irregularity plays an essential role in keeping the high-speed railway safe and stable. Regular maintenance is an important measure to guarantee track smoothness, which has a great influence on the change of track state. This paper introduces the time series anomaly detection model (AD) to detect changes by tracking the difference between the mean values of two sliding time windows. These changes are taken as one of the holidays of the prophet model which is called the AD_ Prophet model. Moreover, this paper proposes a multilevel residual prophet (Re-Prophet) prediction network which can make full use of the information of residual data to predict the results. The final predicted result is the sum of the value of each prediction. To explore the characteristics of the time series, the Hodrick-Prescott filter is used to explore the trend. A multi-month-wise box plot is used to explore the seasonal volatility and the wavelet transform is used to explore the cyclical variation. Based on these characteristics, the fitting model can be chosen. Finally, to verify the high accuracy of the model, the surface irregularities of the high-speed railway are used as the research objects. Compared with conventional prediction algorithms, the proposed model has the smallest prediction deviation and the highest accuracy based on the evaluation indicators such as mean squared error, root mean squared error, mean absolute error, and mean absolute percentage error.

The term “user persona” recently has become more popular and can reflect the characteristics and needs of each user. To analyze the individual characteristics of each passenger in order to better implement targeted transportation policies, a method to mine travel profiles from each passenger based on their smart card transaction records is proposed, mainly mining six labels of passengers: activity, loyalty, whether they are stable commuters, where they live, where they work, and how they prefer to travel. A case study was conducted in the Huitian area of Beijing, the largest community in Asia, which demonstrated the high practicality and accuracy of the passenger profiling method. The results show that 33.06% of passengers were given the stable commuter label and were more likely to have high activity and loyalty labels, whereas 66.94% of other passengers were more likely to have low activity and loyalty labels; 88.9% of passengers were given the residence label and 67.71% of stable commuters received the workplace label, with an accuracy rate of over 90% according to a travel survey. In addition, the application of passenger labels during the design of demand responsive transit (DRT) was discussed to illustrate how to use the labels to improve the efficiency of DRT. The proposed passenger profiling method is applicable to the data mining of passenger travel labels in a simple and accurate way, and can help public transport service providers and researchers to study individual passenger characteristics and provide a theoretical basis for transit network planning and personalization measures.

This paper addresses a key limitation of using existing detection and controller data of signalized intersections to build performance metrics in the automated traffic signal performance measures (ATSPMs) platform. Many intersections were installed with lane-by-lane stopbar detectors for actuated and adaptive signal controllers. Stopbar detector actuations are valuable inputs for signal controllers and intersection performance but are not the ideal advanced detector inputs for several key ATSPM metrics such as arrival on red (AoR) and Purdue coordination diagram (PCD). This paper presents a vehicle trajectory reconstruction algorithm based on shockwave theory to estimate advanced vehicle detections from stopbar arrivals and departures. Model parameters including shockwave speeds, free-flow velocity, acceleration, and deceleration rates were directly measured using spatial-temporal maps (STMaps) generated from roadside closed-circuit television (CCTV) camera video. The initially measured parameters were optimized using the genetic algorithm (GA) that was subsequently validated quantitively and qualitatively. Finally, combining the stopbar detector and signal phase and timing, a new coordination diagram was designed to enable traffic operators to identify mobility patterns and safety-critical events quickly. This research sought to utilize existing data sources to meet performance metric requirements while avoiding intensive investment to upgrade the current infrastructure. The STMap-based method substantially reduces the complexity of obtaining necessary model parameters. The new stopbar-based Rutgers coordination diagram enriches the visualization tools for intersection performance measures from controller and detection systems.

Connected vehicle (CV) data in this paper refer to the in-vehicle telematic data, including trajectories and driving events (e.g., hard braking) collected by vehicle manufacturers when vehicles are moving. Recently manufactured vehicles are equipped with cellular modems and Internet of Things (IoT) devices to collect vehicle data. Such data, after removing personal information, are being redistributed to third-party organizations. Compared to other probe vehicle data, the CV data has a higher penetration rate, ubiquitous coverage, and almost lane-level positioning accuracy. These features pave the road for novel transportation applications in transportation planning and traffic operations. In this paper, we represent a novel framework to estimate the regional link volumes based on the CV data and a deep neural network (DNN) model. The training data are generated according to the link volumes (targeted model output) and the corresponding CV counts (input features) at the same locations. The DNN model’s performance was compared with other estimation methods like linear regression and random forest and showed superior performance. The trained DNN model takes ubiquitous CV counts from other locations to estimate the corresponding link volumes. As a case study, the proposed DNN model was trained with a large training data set derived from CV data and time-dependent link counts collected at over 1,200 locations on freeways in the Dallas Fort Worth, Texas, area. The results reveal good accuracy and robustness.

Cellular signaling data (CSD) is a new data source of intelligent transportation and has great potential in the study of human mobility. However, research recently has shown that the privacy of individual trajectories of travelers can be easily leaked due to their unique travel patterns, which will lead to the disclosure of a traveler’s trajectory privacy. To address this issue, the risk of privacy leakage in a CSD data set was quantitatively measured due to reidentification attacks. To reduce the risk of trajectory privacy leakage, a spatial generalization method based on an improved genetic algorithm (IGA) was proposed. We transformed the base station aggregation problem that meets the k-anonymity requirement into a shortest path problem. The data utility loss after privacy preservation was evaluated based on origin–destination (OD) flow analysis of traffic analysis zones (TAZs). Finally, the relationships between privacy preservation and data utility were revealed. In all of the cases, more privacy protection corresponded to less data utility. The discovered relationships provide useful guidance for data publishers on how to choose the right tradeoff between privacy protection and data utility when publishing or sharing such sensitive data.

Recent data have shown that pedestrian fatalities have increased by 44% from 2010 to 2019. In 2019, 6,590 pedestrians died of traffic crashes, and 20% occurred at intersections, the highest in 30 years. These saddening facts, unfortunately, suggest that walking or biking is even less safe today although driving is safer thanks to numerous efforts in vehicular technologies. Smart transportation is successful only if it provides equitable safety for all road users. Flash yellow arrow (FYA) is a left-turn strategy at signalized intersections in North America. It allows left-turn vehicles to cross when the gaps of opposing through-traffic are perceived as safe. But it cannot separate concurrent crossing pedestrians and left-turn vehicles. To address this issue, a novel dynamic flash yellow arrow (D-FYA) solution is developed using the light detection and ranging (LiDAR)-based tracking technique. It can address the safety concerns in the FYA while recovering the permissive left-turn capacity after the concurrent pedestrians are cleared. Depending on the pedestrian volumes, the corresponding FYA with each cycle will either start as scheduled, be postponed, or be canceled within each cycle. The proposed D-FYA was deployed at an intersection next to the campus of the University of Texas at Arlington, and its real-time D-FYA decisions in the field were verified over 100 traffic signal cycles through simultaneous observation in the field. The proposed D-FYA solution was further evaluated within a traffic signal simulation platform to compare its mobility performance with another two permissive left-turn strategies: (1) protected + permissive left turn (PPLT), and (2) PPLT with minus-pedestrian phase. The experiment results revealed the D-FYA is accurate and adaptive compared with the other two permissive left-turn strategies.

Contending with various recurrent congestion patterns on major freeway corridors has long been a prioritized task of the transportation community. Most of these patterns take place on a freeway interchange’s on-ramp and off-ramp segments where both exiting and merging flows often trigger a series of lane changes, resulting in significant freeway speed reduction and propagation of traffic queues to neighboring surface streets. To concurrently mitigate freeway traffic congestion and minimize any negative impacts on the nearby local arterials, this study has developed an Integrated Real-time Interchange Control (IRIC) system, which maximizes the benefits of both freeway and arterial users by fusing an integrated off-ramp signal control (IOSC) module, an off-ramp queue impact (OQI) model, a lane group-based (LGB) traffic model, and a ramp metering control strategy as well as arterial signal optimization modules. The proposed system with both ramp metering functions and local arterial signal design capabilities can determine whether to implement system-wide optimization or merely the ramp metering based on real-time detected traffic conditions. The evaluation results from experiments have confirmed the effectiveness of the developed real-time interchange control system, especially with respect to its ability of preventing ramp queue overflows to the neighboring arterials and minimizing the off-ramp spillback onto the freeway mainline.

Traffic management strategies have been relying on various congestion prediction methodologies. The prediction accuracy of these methodologies has improved over the years, offering reasonable short-term and midterm predictions of macroscopic traffic measures (i.e., flow, speed, and occupancy/density). Unfortunately, by relying on fixed infrastructure sensors and aggregated data, these prediction methodologies fail to include microscopic traffic flow dynamics in their prediction algorithms. Accordingly, they usually fail to capture the onset of congestion and can only predict the propagation of existing shockwaves. That is, in fact, critical for utilizing effective traffic management strategies because predicting the onset of congestion can significantly help with mitigating it. Addressing this shortcoming in traffic predcition algorithms, this study proposes a deep learning methodology to predict the formation and propagation of traffic shockwaves at the vehicle trajectory level. Assuming the existence of communications between vehicles and infrastructure, the time-space diagram of the study segment serves as the input of the deep neural network, and the output of the network is the predicted propagation of shockwaves on that segment. It is the capability to extract the features embedded in a time-space diagram that allows this methodology to predict the propagation of traffic shockwaves. The proposed approach was tested on both simulation and real-world data, and results show that it can accurately predict shockwave formation and propagation.

A transit network design frequency setting model is proposed to cope with the postpandemic passenger demand. The multiobjective transit network design and frequency setting problem (TNDFSP) seeks to find optimal routes and their associated frequencies to operate public transport services in an urban area. The objective is to redesign the public transport network to minimize passenger costs without incurring massive changes to its former composition. The proposed TNDFSP model includes a route generation algorithm (RGA) that generates newlines in addition to the existing lines to serve the most demanding trips, and passenger assignment (PA) and frequency setting (FS) mixed-integer programming models that distribute the demand through the modified bus network and set the optimal number of buses for each line. Computational experiments were conducted on a test network and the network comprising the Royal Borough of Kensington and Chelsea in London.

Road transportation is a critical mode of transportation because of its mobility characteristics. Road construction work zones (WZ) are widespread on roads because of increased travel demand. Thus, it is necessary to study traffic safety for WZs and regular sections. In this study, traffic safety is analyzed in terms of the conflict probability at selected roads with WZ and without work zone (WWZ) sections along the same road. Vehicular trajectory data for three traffic flow levels (free flow, near capacity, and congestion) were extracted using a newly developed machine learning–based semiautomated trajectory extractor tool. The derived trajectory data from both sections were used to identify the leader-follower vehicle pairs using MATLAB. Two surrogate safety measures were estimated to compute the rear-end conflicts: time-to-collision and deceleration rate to avoid a collision. Hence, the variation of the rear-end conflicts was further examined using the generalized extreme value theory. It is found that the conflict probability is more in the WWZ section than the WZ section, which may be attributed to the variation in speed and acceleration. It is also found that the conflict probability for the WWZ and WZ sections is reduced as the angle between the two vehicles increases. The study results should help highway authorities to implement suitable safety measures to reduce conflicts and crashes in the work zones.