Traffic Engineers Research Articles

Modeling and Forecasting Freight Delivery for Blocked and Unblocked Streets through Machine Learning Approaches and Analysed by Microsimulation Model

Deliveries of freight on signalized city roads often lead to lane obstructions, contributing to urban traffic congestion. This issue has garnered increasing attention as traffic engineers and city planners seek sustainable solutions to meet growing demand while working within the constraints of limited road capacity. The primary objective of this study is to assess a model designed to quantify the impact of freight deliveries on both the capacity and delay experienced on signalized city roads in Ahmedabad. The model is structured similarly to the approach outlined in the Highway Capacity Manual (HCM2010). The aim of this research is to provide insights into the utilization of these analytical tools in the context of urban freight delivery policies. This research explores the estimation of delay and vehicle capacity, accounting for various factors such as delivery locations, durations, and their distinct effects on different lanes. To predict vehicle capacity and estimate delays, this study employed machine learning techniques, specifically Support Vector Machines and Artificial Neural Networks. The results demonstrate a high level of agreement between our predictions and actual observations.

Crossing intersections: A tool for investigating road user pathways

The pathways used by cyclists, pedestrians, and users of micromobility to cross intersections do not always align with those planned by traffic engineers. Observing actual usage patterns could lead to a better understanding of the tactical behavior of users of active and micromobility, allowing planners and engineers to create urban environments specifically for these road users. An open-source Python tool is introduced that uses clustering to automatically identify the forms of pathways used by road users. The tool was used to cluster trajectories from five intersections in Germany. The exemplar of each cluster is selected to represent the average shape of each pathway type. The open-source Python tool RoadUserPathways is introduced, the case studies are examined and use cases are presented.

Automatic Calibration of Microscopic Traffic Simulation Models Using Artificial Neural Networks

Traffic simulations are valuable tools for urban mobility planning and operation, particularly in large cities. Simulation-based microscopic models have enabled traffic engineers to understand local transit and transport behaviors more deeply and manage urban mobility. However, for the simulations to be effective, the transport network and user behavior parameters must be calibrated to mirror real scenarios. In general, calibration is performed manually by traffic engineers who use their knowledge and experience to adjust the parameters of the simulator. Unfortunately, there is still no systematic and automatic process for calibrating traffic simulation networks, although some methods have been proposed in the literature. This study proposes a methodology that facilitates the calibration process, where an artificial neural network (ANN) is trained to learn the behavior of the transport network of interest. The ANN used is the Multi-Layer Perceptron (MLP), trained with back-propagation methods. Based on this learning procedure, the neural network can select the optimized values of the simulation parameters that best mimic the traffic conditions of interest. Experiments considered two microscopic models of traffic and two psychophysical models (Wiedemann 74 and Wiedemann 99). The microscopic traffic models are located in the metropolitan region of São Paulo, Brazil. Moreover, we tested the different configurations of the MLP (layers and numbers of neurons) as well as several variations of the backpropagation training method: Stochastic Gradient Descent (SGD), Adam, Adagrad, Adadelta, Adamax, and Nadam. The results of the experiments show that the proposed methodology is accurate and efficient, leading to calibration with a correlation coefficient greater than 0.8, when the calibrated parameters generate more visible effects on the road network, such as travel times, vehicle counts, and average speeds. For the psychophysical parameters, in the most simplified model (W74), the correlation coefficient was greater than 0.7. The advantage of using ANN for the automatic calibration of simulation parameters is that it allows traffic engineers to carry out comprehensive studies on a large number of future scenarios, such as at different times of the day, as well as on different days of the week and months of the year.

Resilience Assessment Framework for an Urban Road Network Subjected to Disruptions

This research paper introduces a novel framework for assessing the resilience of urban transport networks. Traditional approaches to quantify resilience primarily focus on evaluating resilience based on the accessibility and vulnerability of network links. However, this study proposes a different approach by considering the dynamic role of links within the overall network, including their ability to accommodate additional traffic demand from other links. To accomplish this, a comprehensive scanning network approach is implemented using an iterative traffic assignment performed. The framework is applied to the urban transport network of Karachi, Pakistan. The model incorporates parameters for calibrations derived from volume delay functions, specifically alpha and beta, which are obtained from actual traffic travel times during free-flow and peak hours. The accuracy of the model is validated by comparing observed traffic counts at various locations with the simulated traffic volumes, resulting in an average error of 9% and a standard deviation of 7.4%. The study's findings reveal significant losses within the network and identify critical links that are particularly susceptible to disruptions. The analysis provides valuable insights into delays experienced by travellers and the additional distances travelled when specific links are closed. Furthermore, the framework enables the evaluation of the overall impacts of disruptions at the network level. The proposed methodology will allow traffic engineers and decision-makers to effectively plan rerouted through alternative links during disruptions.

An internet of things prototype to quantify the lateral wheel path distributions of local roads in New Zealand

ABSTRACT The actual lateral wheel path distribution (LWD) on roads differs from the design assumptions. This leads to premature deterioration of chip-sealed pavements in the form of flushing and ravelling resulting in subsequent maintenance costs. The location of LWD is key to calibrating the variable-rate bitumen spray bar used to minimise flushing and ravelling during chip seal construction. Current methods used to estimate the location of LWD rely on visual inspection. This is subjective and reliant on individual expertise. This study aims to automate obtaining LWD by developing an Internet of Things (IoT) prototype to locate the LWDs. This prototype was tested by collecting LWDs from 22 locations in Christchurch, New Zealand. Results show that narrow roads exhibit a concentrated wheel path distribution. Specifically, increasing single lane carriageway width leads to a significant rise in the standard deviation (spread) and position of the LWDs among 22 locations investigated. Cycle lanes or parking bays significantly influence the location and distribution of LWD. There is no clear link between road camber and LWD. The findings can be used to improve the calibration of the variable-rate bitumen spray bar and inform pavement and traffic engineers to enhance construction and maintenance practices. .

A Conflict-Based Safety Diagnosis of SCI Roundabouts Using a Surrogate Safety Measure Model

Recently, the surrogate safety assessment model has been considered for the safety performance analysis of roundabouts. This model can perform a detailed analysis of conflicts based on the trajectory data of vehicles inside the roundabout. The Separated Central Island (SCI) roundabout, as one of the roundabouts with distinct geometrical characteristics, holds the potential for safety evaluation using the conflict-based method. Therefore, in this study, a safety diagnosis of an SCI roundabout was conducted for the first time. In this study, a safety diagnosis procedure for SCI roundabout was first defined; for this purpose, vehicle trajectory data were obtained using an Unnamed Aerial Vehicle (UAV) and then entered into Surrogate Safety Measure Model software (SSAM 3.0). A case study was conducted on the only SCI roundabout in Iran, located in Mashhad. The parameters of Time-to-Collision (TTC), maximum speed difference of two vehicles involved in a collision (MaxΔV), and maximum speed of vehicles in collision (MaxS) were utilized to determine the type and severity of conflicts and risk analysis. The results of roundabout risk analysis showed that the severity level of conflicts is mainly of the injury type and that the lowest severity of conflicts is related to fatality. In addition, the highest frequency of injury conflicts is associated with lane-change conflict and the lowest frequency is linked to rear-end conflict. The highest and lowest frequency of damage conflicts are related to crossing and lane-change conflicts, respectively. After overall risk scoring, the severity level of conflicts is mainly related to injury type and the lowest severity of conflicts is associated with fatality; 31% of the total conflicts obtained are of the damaging type, and 69% are of the injury type. Finally, comparing the results of the conflict data with the 8-year crash data in such roundabouts confirms that in the absence of crash data in such roundabouts traffic engineers can use the roundabout analysis based on this study to predict the safety situation of such roundabouts before implementing engineering processes.

A novel road traffic flow prediction model using hybrid Particle Swarm Optimization (PSO) and Radial Basis Function Neural Network (RBFNN)

<p>Traffic congestion is a major problem in urban areas, leading to increased travel time, air pollution, and fuel consumption. Road impedance function, which describes the relationship between traffic status and travel time, plays an important role in predicting travel time and managing traffic flow. Traditional methods for estimating road impedance function rely on manual calibration and may have limitations in reflecting the complexity of traffic patterns. To address these challenges, researchers have proposed various machine learning models for predicting travel time and road impedance function. In this paper, a hybrid particle swarm optimization—radial basis function neural network model is proposed for improving the accuracy of the road impedance function. The model takes into consideration various vehicle types and is validated using travel time data collected from a road section in Huai’an City, China. The effectiveness of the proposed model is compared with the traditional road impedance function calibrated by nonlinear regression. The experimental results indicate that the Mean Relative Error (MRE) of PSORBFNN is increased by 3.89% and 6.28% respectively when compared with DPNR training samples and validation samples. When compared with DPPSO training and validation samples, the MRE of PSORBFNN is increased by 2.87% and 3.3% respectively. These findings suggest that the proposed model could guide and assist traffic engineers and practitioners in predicting travel time on road sections with improved accuracy.</p>

Investigation of factors affecting turn signal usage at modern roundabouts in State of Qatar

ABSTRACT Drivers are expected to use turn signals to indicate other drivers of their intentions of taking turns, lane changing, merging or diverging indicating safe driving behaviour. This study investigated the rates of turn signal usage at multilane urban roundabouts along with factors affecting the usage. The results from data collected from four roundabouts in Doha city showed that the drivers have higher tendency of using turn signals while exiting the roundabout rather than while entering the roundabout. Further, turn signal usage at entry and exit of the roundabouts was affected by turn type, roundabout type, correct lane, vehicle type, age, type of clothes and distraction. Additionally, using correct lane affected turn signal usage only while exiting the roundabouts. The binary logistic regression models showed that the drivers wearing formal clothes, taking left or U-turn, using correct lane, driving without distraction, and traversing a roundabout located in recreation and commercial land use had higher odds of turn signal usage than their counterparts. The lower turn signal usage rates indicate that more emphasis should be provided on educating drivers highlighting safety implications. Further, traffic engineers can suggest additional strategies to improve turn signal usage based on the results. These findings can be applicable to other GCC countries exhibiting similar characteristics.

Accuracy of critical gap estimates

The critical gap is an important parameter in capacity models of unsignalized intersections. Although there are many methods for critical gap estimation, there is currently no approach or method available to determine how precise a calculated estimate is, i.e. how close it is to the true mean critical gap. As a result, traffic engineers and researchers risk unknowingly working with imprecise estimates, which can lead to inaccurate or unreliable capacity analyses of unsignalized intersections. The purpose of this paper is to address this problem by providing new methods for measuring the accuracy of critical gap estimates. We focus on estimates obtained by the maximum likelihood method (using the so-called Fisher information), but a method for measuring the accuracy of arbitrary estimation methods (the Bootstrap method) is also provided. The paper shows how the proposed methods help to assess the quality of critical gap estimates, detect imprecise estimates, and provide guidance on how many observations should be made for the calculated values to be sufficiently precise. Specifically, we recommend that at least 100 observations should be made to generally obtain reliable estimates. For ease of use, an Excel spreadsheet implementing the proposed methods is provided.

Between traffic and walking discourse: Pedestrians in the traffic machine, hints from the Estonian case

Walking has attracted increasing interest in the social sciences and humanities. Yet only rarely have studies in those disciplines considered the subject of traffic, leaving governing and regulations to traffic specialists and engineers. Primarily using conceptual reflections but intersecting these with an analysis of historical traffic regulations in Estonia as they are related to contemporary framings, this article introduces the pedestrian as an element of a traffic system who is governed to control her body in legally prescribed ways. Directing attention also to language, as the vignette from Estonia shows the existence of a particular word “liikleja” for the ways in which pedestrians are seen as being subject to the traffic system as disciplined elements of the traffic machine. The article describes such perspective as “traffic discourse” which is contrasted with the more holistic view of pedestrians present in walking studies (“walking discourse”).

Analysis of Model Predictive Intersection Control for Autonomous Vehicles

Autonomous vehicles are in the main focus for automotive companies and urban traffic engineers as well. As their penetration rate in traffic becomes more and more pronounced due to improvement in sensor technologies and the corresponding infrastructure, new methods for autonomous vehicle controls become a necessity. For instance, autonomous vehicles can improve the performance of urban traffic and prevent the formation of congestions with the usage of Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication based control methods. One of the key area for improvement is centralized intersection control for autonomous vehicles, by which traveling times can be reduced and efficiency of traffic flow can be improved, while safety of passengers can be guaranteed through constraints built in the centralized design. The paper presents the analysis of a Model Predictive Control (MPC) method for the coordination of autonomous vehicles at intersections by comparing it with an offline constraint optimization considering time and energy optimal intervention of vehicles. The analysis has been evaluated in high-fidelity simulation environment CarSim, where the speed trajectories, traveling times and energy consumptions have been compared for the different methods. The simulations show that the proposed time-optimal MPC intersection control method results in similar traveling times of that given by the time-optimal offline constraint optimization, while the energy optimal optimization re-quires significantly more time for the autonomous vehicle to achieve. Due to the possibility of a congestion forming in the latter case, the proposed centralized MPC method is more applicable in real traffic scenarios.

Supervised machine learning for jamming transition in traffic flow with fluctuations in acceleration and braking

Jamming transition in traffic flow refers to the sudden transition from a free-flowing state to a jammed state as the traffic density increases. This transition is of great interest to traffic engineers and physicists, as it can have significant implications for traffic safety, efficiency, traffic management, and urban planning. Homogeneous car following models is a popular framework used to study the jamming transition phenomenon. The mathematical structure of the problem is governed by the classical Lorenz system to consider the fluctuational effects. The analytical solution of such nonlinear oscillatory differential equations does not exist. Therefore, this study aims to utilize the machine learning approach with the optimization technique that could be used to fine-tune the weights/ parameters of a neural network model to predict the accurate and reliable solutions for the jamming transition in traffic flow. The headway deviations have been studied by considering the multiple scenarios based on the acceleration and braking of the vehicle.

Distributed Self-Organizing Control of CAVs Between Multiple Adjacent-Ramps

Traffic self-organizing is controlled autonomously by rules that rely on adaptation to local variations in traffic state and enable effective coordination of the vehicular traffic at a network level. Combined self-organizing network with intelligent transportation, we proposed a distributed self-organizing method for Connected and Autonomous Vehicles (CAVs), which aimed to improve the efficiency and safety between Multiple Adjacent-Ramps (multi-ARs). To make the mainline formation more stable, the speed of ramp vehicles was adjusted to ensure suitable speed and headway of the mainline formation. In the test, the multi-ARs in the East Ring Interchange on the Inner Ring Express in Chongqing was selected to collect the initial data sample by the drones and fixed-point cameras. Under the respective scenarios of conventional driving and intelligent networks, the Python, SUMO, and TraCI were adopted to run simulations and validate the proposed model. Results showed that our model could keep Time to Conflict (TTC) above 1.4s, reduce the average delay by 34.22%, reduce the lane-changing times by 28.07%, reduce single lane occupancy to 8% and improve average speed by 3.68% of multi-ARs. To verify the applicability of the proposed model, experiments were carried out under different traffic volumes, demonstrating the relevance of the proposed method for medium-to-high-density traffic flows. It can provide a basis for traffic engineers and policymakers to maintain the stable development of the urban expressways and ensure the overall operation quality of the multi-ARs.

Machine learning-based multi-target regression to effectively predict turning movements at signalized intersections

Effective prediction of turning movement counts at intersections through efficient and accurate methods is essential and needed for various applications. Commonly predictive methods require extensive data collection, calibration, and modeling efforts to estimate turning movements. In this study, three models were proposed to estimate turning movements at signalized intersections using approach volumes. Two sets of data from the United States and Canada were obtained to develop and test the proposed models. Machine learning-based regression models, including random forest regressor (RFR) and multioutput regressor (MOR) in addition to an artificial neural network (ANN) model, were developed and trained to analyze the relationship between approach volumes and corresponding turning movements. Multiple evaluation measurements were utilized to compare the models. All models produced satisfactory results. The RFR regression model outperformed the MOR model. However, the ANN model had the best performance when compared to the other models. The proposed models provide traffic engineers and planners with reliable and fast methods to estimate turning movements.

Intersection capacity adjustments considering different market penetration rates of connected and automated vehicles

ABSTRACT To better prepare and guide both intersection planning and operations under different market penetration rates (MPRs) of connected and autonomous vehicles (CAVs) and traffic demands, this research estimates both the lane-level and intersection-level capacity. On the lane-level, adjustment factors for saturation headway and saturation traffic flow rate for each lane under different MPRs of CAVs are investigated. On the intersection-level, the maximum throughput function for different MPRs of CAVs is calibrated. With 100% CAVs, the saturation headways for the exclusive through lane, exclusive left-turn lane, and shared-right-and-through lane decrease by 55.8%, 48.9%, and 42.4%, respectively. The maximum throughput of the intersection with 100% CAVs increases by 70% compared to the scenario with only human driving vehicles (HDVs). Moreover, the maximum throughput increases rapidly after 60% MPRs of CAVs. The research could provide a solid reference for traffic engineers and planners in calculating the intersection capacity under different MPRs of CAVs.

The Effect of the Number of Right-Turn and Left-Turn Lanes on the Performance of Undersaturated Signalized Intersections

Vehicle traffic flow channelizing can significantly contribute to fewer collisions and smooth traffic flow in isolated signalized intersection design. Therefore, this study investigates the effect of changes in the number of right-turn lanes with splitter islands and left-turn storage bays for different volume approaches on intersection performance. This study aims to provide an accurate analysis to estimate the effect of such changes on signalized intersections, which can improve intersection performance. Furthermore, the delay parameter derived from HCM2016 and simulation was evaluated for different scenarios. For this purpose, simulation was performed using Synchro software. In this study, a symmetrical, 90-degree, undersaturated (volume per capacity lower than 1, V/C < 1), and four-leg intersection were considered according to AASHTO Green Book 2018 suggestions. HCM2016 results indicated the delay parameter was less sensitive to the undersaturated condition than the right-turn volume variation for both single and dual-lane bays. However, the simulation results indicated that the delay parameter was not constant, depending on the number of right-turn lanes and volumes. On the other hand, there was a significant difference in delay parameters between the single and dual left-turn lanes for both simulation and HCM cases estimated 77.4% and 59.7%, respectively. Overall, this study can provide a vision for traffic engineers to modify the geometry of the four-leg signalized intersections, if the right-turn or left-turn demand volume of an undersaturated signalized intersection is larger than the through approach.

Prediction of Risk Factors Influencing Severity Level of Traffic Accidents Using Artificial Intelligence

Road safety is of major interest for highway and traffic engineers worldwide. In Jordan, road networks have recently displayed relatively high traffic volumes, specifically in urban centers and in the Central Business District (CBD) areas of major cities. Irbid is one of the major cities suffering from serious traffic accidents problems that should have received more attention from decision makers. Based on data of 94,356 accidents occurred over a five-year period (from 2013 to 2017), Artificial Neural Network (ANN) and Support Vector Machine (SVM) techniques have been employed to predict and classify traffic accidents in this city. ANN has been used to model the relationship between driver injury severity and traffic accident factors, such as age and gender of drivers, type and faults of vehicles, weather conditions and reasons of accidents. The structure of the used ANN model has been one input layer of 6 neurons, with one hidden layer of 15 neurons and one output layer of 3 neurons representing severity level. The ANN model has showed high correlation based on the high value of correlation coefficient R=0.87. The used ANN model has exhibited better results in predicting new accidents with MSE equal to 0.05, compared with SVM model with MSE of 0.09. Sensitivity analysis has been carried out on the trained neural network to identify the importance of crash-related factors. The traffic accident data have been used to build the classifier, using SVM. The overall model classification performance has been 90.4%, which accounts for the circumstances under which drivers are more likely to be killed or injured in a vehicle accident. It has been concluded that the comprehensive performance of the SVM model is better than the ANN model for traffic accidents classification.

Control and allocation at an intersection with traffic lights

Managing road infrastructure in highly urbanized areas during the Industry 4.0 era poses a significant technological challenge for traffic engineers. Steadily growing traffic volume, both of motor vehicles and pedestrians, cyclists or scooters, necessitates the development of advanced traffic control algorithms and systems. Improvement of the efficiency and capacity of intersections without any intervention into the physical structure of roadways is one of the two aspects of today's traffic engineering. In this paper, however, emphasis is placed on the aspect of coordination of conflicting flows of motorized and pedestrian traffic. Consideration is given to ensuring that each traffic light phase at an intersection is maximally utilized while maintaining the required safety factor. The empirical part of the study consists in identifying inefficiencies in the traffic light cycle. The analysis of traffic phases has revealed that there were certain periods in the cycle during which pedestrians were given the right of way while, in fact, it was not necessary. The study proposes a solution to optimize the green signal timing for conflicting vehicular and pedestrian flows in real-time. By eliminating delays in traffic signal phases as a result of not having to give a green signal for pedestrians, the proposed solution will improve the efficiency of urban networks with respect to an aspect whose potential has not been exploited so far. Coordination and adaptation of each light phase to the most effective extent is a feasible answer to the needs and requirements of modern urban areas.

A MULTI-OBJECTIVE OPTIMIZATION FRAMEWORK FOR TRAFFIC SIGNAL DESIGN

Traffic signal design requires experienced traffic engineers to decide on the phasing plans and then timing plans can be optimized. However, the available guidance provides general recommendations for phasing planning; while, commercial software optimizes the timing plans and corresponding operation performance (e.g., best performance index). Furthermore, the performance index excludes safety factors. This research aims to develop a framework that can deliver a phasing plan by enumerating all possible solutions and then selecting the phase plan that achieves the objectives. These objectives are to minimize the severity, average delay, and queue length. A program (i.e., software) has been developed to perform these stages and deliver a sorted list of phase plans based on the assessment criteria considering different traffic patterns, different intersection lane configurations, and operation types for left and right turns. A validation exercise was performed to assess the effectiveness and practicality of the developed software. It includes designing a phase plan for fifteen American, Canadian, and Chinese intersections and comparing them against their actual phasing design in terms of safety, average delay, and queue length. The results show that on average, the safety level is improved by 19%; however, the delay and queue length increased by 33% and 13%, respectively. The results encourage extending the proposed framework to consider the phasing sequence and coordination in future work. Finally, it is found that the framework provides practical phasing plans in terms of safety and other operational aspects. The results encourage extending the proposed framework to consider the phasing sequence and coordination in future work. Keywords: Traffic Signal Optimization, Phasing Plan, Highway Safety Manual, Queue Length DOI： https://doi.org/10.35741/issn.0258-2724.58.1.37

3-step calibration process of a microscopic traffic flow simulation platform for mixed traffic scenarios

Microscopic traffic flow simulation is used in multiple fields nowadays, resulting in varying objectives and requirements regarding model calibration. Traffic engineers analyse the impact on traffic flow using simulation tools and in automobile industry simulation is used for developing and analysing automated driving functions and their functionalities. We developed a simulation platform with pre-defined applications, allowing operators to test the impact of automated driving on traffic flow by varying mixed traffic scenarios. As calibration of the base case for different domains is a time-consuming process, we present in this paper a 3-step calibration process, which allows efficient model calibration by focusing on macroscopic traffic flow dynamics as well as microscopic driving behaviour.