Skip to main content

2024 | Book

Traffic and Granular Flow '22

Editors: K. Ramachandra Rao, Armin  Seyfried , Andreas Schadschneider

Publisher: Springer Nature Singapore

Book Series : Lecture Notes in Civil Engineering


About this book

This book gathers contributions on a variety of flowing collective systems. While primarily focusing on pedestrian dynamics, it also reflects the latest developments in areas such as vehicular traffic and granular flows and addresses related emerging topics such as self-propelled particles, data transport, swarm behaviour, intercellular transport, and individual interactions to complex systems. Combining fundamental research and practical applications in the various fields discussed, the book offers a valuable asset for researchers and professionals in areas such as civil and transportation engineering, mechanical engineering, electrical engineering, physics, computer science, and mathematics.

Table of Contents


Pedestrian Dynamics

Collective Traffic of Agents That Remember

Traffic and pedestrian systems consist of human collectives where agents are intelligent and capable of processing available information, to perform tactical manoeuvres that can potentially increase their movement efficiency. In this study, we introduce a social force model for agents that possess memory. Information of the agent’s past affects the agent’s instantaneous movement in order to swiftly take the agent towards its desired state. We show how the presence of memory is akin to an agent performing a proportional–integral control to achieve its desired state. The longer the agent remembers and the more impact the memory has on its motion, better is the movement of an isolated agent in terms of achieving its desired state. However, when in a collective, the interactions between the agents lead to non-monotonic effect of memory on the traffic dynamics. A group of agents with memory exiting through a narrow door exhibit more clogging with memory than without it. We also show that a very large amount of memory results in variation in the memory force experienced by agents in the system at any time, which reduces the propensity to form clogs and leads to efficient movement.

M. Danny Raj, Arvind Nayak
“Nudging” Crowds: When It Works, When It Doesn’t and Why

In recent years, there has been as substantial technological improvement in pedestrian detection and numerical modeling. Yet, crowd steering is still based on constructional changes or on-site guidance with little automation. In this work, we investigate the possibility of using environmental stimuli to modify (collective) behavior or people. Three different scenarios are considered where steering method, interaction time (with the surrounding environment) and crowd density are changed. Results show that simple changes in land- and soundscape are not sufficient to modify human route choice in a familiar environment such when entering an office building. However, using supervised experiments we showed that when crowd density is sufficiently high, interaction time login enough and the context “neutral”, it is possible to “nudge” people into a more efficient motion. The outcomes of this work may help in the development of steering systems to be used in sparse crowds with minimal constructional intervention at the scope to reduce congestion and delay the occurrence of dangerous situations.

Claudio Feliciani, Sakurako Tanida, Masahiro Furukawa, Hisashi Murakami, Xiaolu Jia, Dražen Brščić, Katsuhiro Nishinari
Revisiting the Theoretical Basis of Agent-Based Models for Pedestrian Dynamics

Robust agent-based models for pedestrian dynamics, which can predict the motion of pedestrians in various situations without specific adjustment of the model or its parameters, are highly desirable. But the modeller’s task is challenging, in part because it mingles different types of processes (cognitive and mechanical ones) and different levels of description (global path planning and local navigation). We argue that the articulations between these processers or levels are not given sufficient attention in many current modelling frameworks and that this deficiency hampers the effectiveness of these models. Conversely, if a decision-making layer and a mechanical one are adequately distinguished, the former controlling the desired velocity that enters the latter, and if local navigation is not guided solely by intermediate way-points towards the target, but by broader spatial information (e.g., a floor field), the greater robustness can be achieved. This is illustrated with the ANDA model, recently proposed based on such considerations, which was found to reproduce a remarkably wide range of crowd scenarios with a single set of intrinsic parameters.

Iñaki EcheverrÍa-Huarte, Alexandre Nicolas
An Emergency Evacuation Model for Avoiding High Nuclide Concentration Areas in Nuclear Accident

This paper proposes a nuclear accident emergency evacuation model based on nuclide concentration field and traffic flow network model. Gaussian puff model, which describes the instantaneous concentration of nuclide, is widely used to predict the diffusion of radionuclides in nuclear accidents. We adopt a modified Gaussian puff model to preserve the time characteristics of nuclide diffusion, which considers the effects of dry deposition (gravity deposition), wet deposition (rain wash), and nuclide decay on the concentration distribution. An A-star algorithm based on nuclide concentration and traffic pressure cost is proposed to formulate the optimal path planning under nuclear accident. With the expansion of the nuclide concentration field, more and more paths cannot be safely passed through, and evacuation vehicles will also cause congestion on the path. The evacuation model we proposed will regularly update the optimal route according to road traffic conditions and nuclide spatial distribution, so as to obtain an optimized evacuation strategy. This study takes a real road network around the nuclear power plant as the application scenario of evacuation simulation. The computation time of the evacuation model is short, so it can respond quickly in the event of a nuclear emergency. The results show that the evacuation strategy formulated by the model can ensure good evacuation efficiency and avoid areas with high nuclide concentrations during the evacuation process, thereby reducing the radiation exposure of personnel.

Zhonghao Zhan, Weiguo Song, Jun Zhang, Chuanli Huang
Shoulder Rotation Measurement in Camera and 3D Motion Capturing Data

The individual movement of pedestrians and their body parts, as for example shoulders, is of great interest to understand body movement and interactions and thus to improve pedestrian models. Nearly all laboratory experiments in pedestrian dynamics use camera data to obtain trajectories. A perpendicular top view of the camera does not only allow to extract the head position but also data of upper body segments. The detection is more reliable if shoulders are tagged with markers and for low densities of people. In this study a head-shoulder model is used to assign coloured shoulder markers to a person. The location of a marker is predicted by taking head position, basic body dimensions, movement direction and camera angle into account. It is implemented as a new feature in the software PeTrack. This paper shows a comparison of shoulder rotation measurements obtained from 3D motion capturing systems (Xsens) with those from camera data using the newly introduced model and detection technique. Detection rates and limits of the camera-based rotation measurement are shown and implications are given for the future application at high densities in crowds.

Ann Katrin Boomers, Maik Boltes
Methods of Density Estimation for Pedestrians Moving in Small Groups Without a Spatial Boundary

For a group of pedestrians without any spatial boundaries, the methods of density estimation is a wide area of research. Besides, there is a specific difficulty when the density along one given pedestrian trajectory is needed in order to plot an 'individual-based' fundamental diagram. We illustrate why several methods become ill-denied in this case. We then turn to the widely used Voronoi-cell based density estimate. We show that for a typical situation of crossing flows of pedestrians, Voronoi method has to be adapted to the small sample size. We conclude with general remarks about the meaning of density measurements in such context.

Pratik Mullick, Cecile Appert-Rolland, William H. Warren, Julien Pettré
A Review of Entropy-Based Studies on Crowd Behavior and Risk Analysis

Understanding crowd behavior has become imperative as the number of mass gatherings is increasing. Many factors, such as overcrowding, venue deficiencies, rumors, accidents (such as fires, etc.), and inadequate crowd management practices, can lead to serious crowd-related severe that have resulted in injuries and fatalities. Several studies have been conducted to understand crowd behavior and predict risk conditions using different methods. The methods used to detect the crowd anomaly behavior can be classified into two general categories; Implicit Method (relies on expert opinions on several factors and needs human intervention in the model) and Explicit Method (Assessment of crowd risk situations done by the model itself, requires less human intervention in the model). Entropy, one of such explicit methods, has been widely adopted to analyze crowd behavior and predict crowd risk situations in recent years. The purpose of this study is to review the existing literature on entropy-based crowd risk prediction models and identify research gaps so that more in-depth studies on crowd management and risk assessment can be conducted.

Kiran Naik, Gayathri Harihara Subramanian, Ashish Verma
Understanding the Difference in Social Group Behaviour of a Spiritually Motivated Crowd and a General Crowd

Over time, numerous studies have been conducted on pedestrian behaviour to improve the fidelity of pedestrian models considering pedestrians as individual entities. The participation of social groups is much higher than individuals in mass religious gatherings, which calls for studies focusing on understanding group behaviour in a crowd. Data was collected at two different settings (a) in Kumbh Mela 2016 representing mass religious gatherings and (b) Open day event held at Indian Institute of Science campus representing a regular urban setting. Trajectories of groups were extracted, and spatial formation of different group sizes were plotted. It was observed that group size 3 formed a linear or V-pattern and groups size 4 and 5 formed asymmetric irregular polygons. The area occupancy of groups and their average walking speeds were also calculated for both datasets, and it was observed that despite Kumbh Mela groups occupying lesser area, the average walking speed is higher than the groups in Open day. Looking at these group behaviour characteristics, this paper tries to uncover how group behaviour in mass religious gatherings is different from a low or moderate density setting and whether or not, there is a need for separate walking behaviour parameters.

Gayathri Harihara Subramanian, Ankit Rai, Ashish Verma
Mass Evacuation Planning Based on Mean Field Games Theory

This study propose a dynamic framework for a mass evacuation in a large-scale urban area. The evacuation plan includes three main decisions: Shelter (destination), departure time, and route. The first decision can be formulated as a shelter allocation problem, while the two latter decisions can be addressed by traffic assignment models. Unlike previously proposed dynamic models that solve shelter allocation and traffic assignment problems separately at the microscopic level, the proposed model addresses the population evacuation at the network level for a large but finite population of agents. We formulate the model based on an extension of infinite population game theory called Mean Field Games. The proposed dynamical system has two sets of equations: conservation equations for the traffic flow dynamics and Hamilton–Jacobi–Bellman (HJB) equation for the trajectory of the particles. The solution of the model is derived by solving the corresponding HJB with respect to the conservation equations, which is equivalent to solving a fixed-point problem. A heuristic algorithm is proposed to find the fixed point.

Negin Alisoltani, Mostafa Ameli, Megan M. Khoshyaran, Jean-Patrick Lebacque
Experimental Study of Bidirectional Pedestrian Flow in a Corridor with Certain Height Constraint

Bidirectional flow is believed to be the main cause of trampling disaster in mass gathering events. To control such kind of disasters, deeper insights into the pedestrian movement features could help. However, bidirectional pedestrians flow of stoop walking has barely been studied. Such form of movement can usually be found when firefighters entered the building to rescue the evacuees who escaped from dangerous zone. Therefore, a series of bidirectional flow experiments with a constraint height of 1.4 m were conducted in order to investigate this movement features. The lanes formation, relationship between speed and density, specific flow and density were analyzed. Results indicated that the numbers of lanes gradually increase with the global density increase. In speed-density relation, when the density is less than 0.8 ped/m−2, the free speeds of bidirectional pedestrian flow is larger when compared with unidirectional pedestrian flow. In specific flow and density relationship, the specific flow of unidirectional pedestrian flow is larger than that bidirectional pedestrian flow when the density is more than 0.6 ped/m−2.

Shi Dongdong, Chen Juan, Chen Jun, Ma Jian
Crowd Dynamics of a Rural Group in a Mass Religious Gathering: A Case Study of Kumbh Mela - 2016, India

Numerous instances of stampede have been witnessed and many lives have been lost till date in mass religious gatherings. While trying to model the evolution of crowd for planning, design and assessment of facilities, major focus has been on simulating such scenarios by modelling individual behaviour whereas group behaviour and effect of cultural differences have been overlooked to a large extent. This study focusses on understanding the group behaviour of people coming from a rural base in a dense setup like Kumbh Mela in India. Both rural and non-rural groups were observed, and data was used to infer understanding of diversity in group behaviour. The predominant shape of groups and their area occupancy were looked upon to draw inference on representative spatial formation for different group sizes and varying personal space of individuals. It was observed that for all group sizes, the group's cohesion and its shape vary with respect to the prevailing density conditions and proximity-to-destination. This study presents the inferences drawn from this unique empirical data and paves way for further work, which by utilizing the results of this study, proposes to test a constructed simulation against a synthetic simulation by incorporating behavioural attributes of crowd in the model.

Ankit Rai, Gayathri Harihara Subramanian, Ashish Verma
Modeling of Obstacle Avoidance by a Dense Crowd as a Mean-Field Game

In this paper we use a minimal model based on Mean-Field Games (a mathematical framework apt to describe situations where a large number of agents compete strategically) to simulate the scenario where a static dense human crowd is crossed by a cylindrical intruder. After a brief explanation of the mathematics behind it, we compare our model directly against the empirical data collected during a controlled experiment replicating the aforementioned situation. We then summarize the features that make the model adhere so well to the experiment and clarify the anticipation time in this framework.

Matteo Butano, Thibault Bonnemain, Cécile Appert-Rolland, Alexandre Nicolas, Denis Ullmo
Two Types of Bottlenecks in Leisure Facilities: Bottlenecks Caused by Attractiveness and Structural Layout

In this study, we investigate the effect of congestion on visitor behavior in Kaiyukan aquarium in Osaka, Japan. Kaiyukan is one of the largest aquariums in the world, where visitors walk through a fixed one-way aisle, unlike most museums or aquariums, where visitors’ movement is not restricted. We acquire data using Bluetooth receivers displaced throughout Kaiyukan and monitor more than 3000 trajectories of visitors carrying beacons. We perform survival analysis and discover that when the exhibition areas are congested, the time spent by visitors viewing an exhibition decreases. Furthermore, two different causes of congestion are identified: the attractiveness of exhibitions and structural layouts. Our findings can provide insights into the congestion phenomenon and the importance of crowd management for reducing congestion in such leisure facilities.

Riho Kawaguchi, Claudio Feliciani, Daichi Yanagisawa, Shigeto Nozaki, Yukari Abe, Makiko Mita, Katsuhiro Nishinari
Face-Validation of a Route-Choice Module in a Crowd Simulator for Confined Indoor Spaces in Context of the COVID-19 Pandemic

The COVID-19 pandemic has drastically changed the life of most people in the world. Governments have come up with various restrictions and measures to contain virus spreading in public (confined) indoor spaces. Simulation tools can reproduce people behaviour and assess how people react on different intervention measures, and are of high value for the related stakeholders. This research extends an existing microscopic pedestrian simulation model for this purpose, where three behavioural decision (strategic, tactical, and operational) levels are considered and adapted to capture crowd movement dynamics in a new context, i.e., indoor locomotion in tight spaces. To validate the newly developed modules, empirical data is collected in real life. This paper presents the effort of collecting empirical data that describe pedestrian route choice behaviour in one of the typical indoor spaces - restaurants. The face-validation for the corresponding route choice module has been conducted. Our study reveals influential factors (i.e., size, location) and behavioural insights regarding tactical decisions in the chosen confined space.

Yufei Yuan, Martijn Sparnaaij, Winnie Daamen, Dorine Duives
On the Influence of Group Social Interaction on Intrusive Behaviours

Having extensively investigated the influence of social bonding on the spatial dynamics of two-people groups (i.e. dyads), we more recently studied the impact of group social relation on the dynamics of individual pedestrians (i.e. non-groups) in their proximity, and, reciprocally, groups’ reaction to such encounters. In the present work, we extend this analysis to additionally study the effect of the groups’ intensity of social interaction (i.e. talking to each other, performing hand gestures, or maintaining eye contact) in similar situations. specifically, using trajectories of uninstructed pedestrians observed in an ecological setting, we analyse encounters between a dyad annotated with an intensity of interaction ranging from 0 (not interacting) to 3 (strongly interacting) and a non-group coming in the opposite direction. We compute the undisturbed minimum distance between them and compare it to the actual minimum distance. To account for the correlation between the intensity of interaction and the size of a group (i.e. the interpersonal distance between the group’s members), the two distances are normalized by the average size of groups with similar intensities of interaction. In line with our previous findings, we demonstrate that avoidance dynamics is more pronounced for groups with higher levels of interaction, while groups that interact less, or not at all, are more likely to be intruded into.

Adrien Gregorj, Zeynep Yücel, Francesco Zanlungo, Takayuki Kanda
Sound Guidance on Evacuation under Limited Visibility: An Experimental Study

In an emergency, like fire or terroristic attack, visibility reduces due to the power failure or smoke, and the sound guidance could be an effective method to improve the pedestrian evacuation. Thus, in this study, we experimentally investigated the evacuation efficiency under conditions of various sound guidance, with different limited visibility. By the extracted trajectories of pedestrians, we analyzed the temporal-spatial features of the density. The main findings include: (1) With the exit width of 1.2 m and the 20% visibility, the sound guidance improves the evacuation efficiency by 12.6%, and the mean time lapse between two successive evacuees passing through the exit reduces by 22.3%; (2) Under limited visibility conditions the power-law index of the complementary cumulative distribution function (CCDF) for the exit under the sound guidance is always higher than that without sound guidance, implying a more fluent bottleneck flow under sound guidance; (3) The sound guidance has limited effect on the crowd temporal-spatial distribution, while the visibility plays a dominant role.

Tao Li, Zhanbo Sun, Zhijian Fu, Lin Luo, Xudong Zhou
Pedestrian Kernel Density Estimates: The Individual Approach

The pedestrian density evaluation problem can be generally perceived as a sampling exercise. Assuming a density distribution inside the analysed area is already generated, the task its to generate a number characterizing the state of an area (detector estimates) or the surroundings of a pedestrian (individual approach). Application of individual density may be crucial e.g. to calibrate any microscopic model or to measure any interaction between pedestrians. Thus, this contribution deals with the individual concept using a conic kernel to generate the density distribution ensuring a pedestrian blurring with great performance. Then the density in pedestrian surroundings of an arbitrary shape with a specific range for a specific pedestrian is defined using kernel distributions. The influence of the shape of pedestrian surroundings and its size described by parameter r are presented suing quantitative metrics. The following type of surroundings (using a different r) are examined: circle (radius); ellipse (length of semi-axes) and sector (radius, angle) which are rotated in accordance with a direction of a pedestrian movement.

Jana Vacková, Marek Bukáček
Density Dependence of Stripe Formation in a Cross-Flow

Qualitative observations in ecological settings along with theoretical and numerical models suggest that when two different pedestrian streams cross a shared area, stripe-like self-organised structure emerge in order to minimise collisions and facilitate the flow. Although the phenomenon has been known for relatively long time, a systematic and quantitative verification of it through controlled experiment has been performed only recently. In this work we analysed such an experiment in which the geometry was kept fixed while changing density, in order to verify if there is a minimum density for stripe formation, and more in general the dependence on density of the phenomenon. An analysis based on two different observables, namely the angle identifying the position of the first neighbour in the same flow, and an order parameter to identify the direction of the environment presenting the higher regularity (the presence of a stripe) suggests that, although the stripe formation pattern is particularly strong at intermediate densities, the tendency to walk on a diagonal stripe is present also at considerably low densities.

Francesco Zanlungo, Claudio Feliciani, Hisashi Murakami, Zeynep Yücel, Xiaolu Jia, Katsuhiro Nishinari, Takayuki Kanda
Estimation of Pedestrian Crossing Intentions in In-Vehicle Video

In recent years, there has been a lot of research on mathematical models of traffic flow, and various models have been proposed for vehicles and pedestrians. Such mathematical models are often used to simulate traffic congestion and disaster evacuation because they behave close to reality under certain conditions. In such simulations, the destination and travel path states of individual agents are often limited. However, when looking at actual traffic flow, individual vehicles and pedestrians often have a large degree of freedom in their destinations and travel paths. Estimating such an initial state is not easy because it requires consideration of not only physical quantities such as the speed and direction of movement, but also the pedestrian’s intentions, such as which direction they are trying to go. Therefore, this study addresses the inference of the intentions of the persons in the video. In this study, we assume in-vehicle video and classify whether the pedestrian in the video is attempting to cross the roadway or not. Experimental results showed that classification of in-vehicle video datasets using a 3D-extended convolutional neural network is possible with an accuracy of more than 70%. Prediction using multiple frames with the 3D-CNN was also shown to be more accurate than prediction using only one previous frame.

Yuto Oyama, Toshiya Takami
Particle Method for Macroscopic Model of Coupled Pedestrian and Vehicular Traffic Flow

In this work, we develop a model to study the interaction between pedestrians and vehicles in a ‘shared sapce’. A non-local macroscopic pedestrian dynamics model is coupled to a microscopic vehicular traffic model in 1D, through eikonal equations. We highlight the use of a mesh-free particle method to solve the hydro-dynamic equations in Lagrangian form, which can then be easily combined with the kinematic equations. The eikonal equations, which provide the minimal path for pedestrians and vehicles, are solved on a fixed structured grid using a fast marching method. The variable values from the particle cloud are interpolated onto the eikonal grid and vice-versa for computation. We demonstrate the model by solving it numerically in three traffic scenarios; one, at an uncontrolled crossing in a single-lane road, second, at a zebra crossing in a single-lane road and finally, at a crossing with a traffic island in a two-lane road.

Parveena Shamim Abdul Salam, Sudarshan Tiwari, Axel Klar, Subbiah Sundar
Study of Emergency Exit Choice Behaviour at Metro Stations in Fire Evacuation

Emergency evacuation is crucial for any unprecedented events like fire and terrorist attacks inside any underground metro rail station. Understanding emergency route and exit choice behavior of metro rail commuters is vital, which can be done through exit choice modelling. In this study, a realistic choice experiment is conducted to understand the exit choice behavior of commuters during fire emergency evacuation from different metro stations of Delhi metro. Sketchup-3D models are used to make the scenarios more realistic. Participants are asked to choose one exit out of the three given choices in each scenario. The exit choice behaviour in different situation is recorded. The choice data is used to calibrate different choice models (standard logit and mixed logit models). The study finds that mixed logit models (Mixed MNL) describe the choices of the respondents better than the standard logit models (MNL). This study also shows that the incorporation of latent factors improves the model fit to a greater extent. The results of this study can be used to develop or validate the exit choice behavior module of any egress simulation software. The results can also be used to design better evacuation systems for metro stations.

Tarapada Mandal, K. Ramachandra Rao, Geetam Tiwari
A Non-linear Pedestrian Tracker Using Velocity-Adaptive Particle Filter with Trajectory Analysis

Object tracking, pedestrian tracking in particular, has a wide range of applications such as video surveillance and pedestrian dynamics. Tracking by detection is a common paradigm for this task. As for tracking pedestrian in videos, most of the existing researches take lots of effort on constant and linear models to estimate pedestrian’s location. Pedestrian’s motion in real life is more likely to follow nonlinear, non-gaussian and multi-modal system rules since they can turn, stop and move at any time. We propose a non-linear pedestrian tracker, using tracking by detection paradigm and developing a velocity-adaptive particle filter to estimate the posterior probability density, so as to estimate given pedestrian motion and track them in video sequence. By the way, the pre-trained detector and simple association method are used for updating the observation template. Our method can achieve better accuracy and precision compared to other filtering tracking method and can successfully track designated pedestrian moving in the presence of occlusion, changeable light intensity and complex dynamic background practically.

Xin Yuan, Weiguo Song, Yang Cao
Wheelchair and Phone use During Single File Pedestrian Movement

Crowd movement studies often use controlled conditions and participants without disabilities. While this approach is reliable, it does not fully reflect real-world conditions and the influence of physical ability as well as secondary tasks. This study examined how these factors affect the headway-speed relationship in single-file movement, including pedestrians and two wheelchair users. Participants followed an experimenter and sometimes engaged in a secondary task (using a phone). Results showed that wheelchair users maintained a constant speed and kept longer minimal headway compared to pedestrians. The secondary tasks did not affect the headway-speed relationship. More research is needed on the implications of mobility profiles and secondary tasks for pedestrian movement in diverse conditions.

Paul Geoerg, Ann Katrin Boomers, Maxine Berthiaume, Max Kinateder, Maik Boltes
Modelling Pedestrian Collective Dynamics with Port-Hamiltonian Systems

Port-Hamiltonian systems (PHS) are increasingly popular modelling approaches for nonlinear physical systems. In this contribution, we identify a general class of microscopic force-based pedestrian models that can be formulated as a port-Hamiltonian system. The port-Hamiltonian paradigm allows for identification of new fundamental physical modelling components of pedestrian dynamics. The skew-symmetric term specific to the conservative Hamiltonian structure of the PHS corresponds to pedestrian isotropic interaction forces. The dissipation to the input port accounts for the pedestrian’s desired velocity and sensitivity, the input acting in the PHS as a feedback control. Some simulations of counter-flow are performed on a torus. Interestingly, a phase transition from disorder dynamics to self-organising lane formation occurs as the conservative forces become weak relative to the dissipation and control forces. A critical parameter setting for lane formation can then be identified using the Hamiltonian as an order parameter.

Antoine Tordeux, Claudia Totzeck, Sylvain Lassarre, Jean-Patrick Lebacque
Experimental Study of Pedestrian Crossing Mechanism in Crowds

Pedestrian crossing behavior is ubiquitous, while relevant studies are limited. In this paper, a series of supervised experiments are conducted to explore the movement characteristics of pedestrians crossing different crowds. As demonstrated from the trajectories, when pedestrians cross the static crowd, anticipation behavior makes pedestrians cross the fixed cross-channel more frequently. Correspondingly, when pedestrians cross the dynamic crowd, anticipation behavior has a shorter duration, and pedestrians can accelerate or decelerate in advance without being able to plan the route. Moreover, spiral-shaped trajectories, a compulsory movement mechanism similar to the phenomenon of turbulence in crowd disaster resulting from the forces acting from the crowd, were observed among individuals in extremely high-density situations. Studies on these crossing behaviors will help us investigate the mechanisms of crowd turbulence formation.

Jinghui Wang, Wei Lv, Yajuan Jiang
Public-Space Sonification for Pedestrian Trajectory Nudging

Increasing the effectiveness and pervasiveness of crowd management measures is an urgent societal need given the continuous urbanization and the persistent growth of large crowd events. Rendering our environments smart and capable of dynamically nudging the crowd flow would provide an effective solution. This holds especially when a massive deployment of field stewards is impossible or not preferred. Yet, this is an outstanding scientific and technological challenge. Here, we present a proof-of-concept towards the usage of acoustic feedback as a way to smarten our environments and automate pedestrian guidance. We established a living lab experiment in a building at TU/Eindhoven (NL). For about 12 weeks, we tracked pedestrians in real-time and reproduced sounds (piano chords, in a wave field synthesis system) coherent with the pedestrian trajectories. We aimed at having pedestrians detouring to a path different from the “typical” for the area. We compare the effect of the acoustic feedback with the control condition (no sound, in random alternation with the feedback). After a transient “learning” phase, pedestrians appear to unexpectedly act contrarily to the feedback design, and rather detoured in opposite direction. This work substantially extends the analysis previously presented by the same authors in [1]

Alessandro Corbetta, Toros Senan, Lex Wöstemeier, Bart Hengeveld
How do Retail Stores Affect Pedestrian Walking Speed: An Empirical Observation

Pedestrian studies in retail areas are critical for comfort and convenience in transportation facility designs. But existing literature lacks detailed empirical observations that focus on pedestrian speed variations and their mechanisms in front of stores. This paper bridges this gap by analyzing 1193 pedestrian trajectories in front of a convenience store located in a metro station. The results show that the store imposes a non-uniform slowing effect on the pedestrian flow. The spatial distribution and the lower walking speed of consumers and gazing pedestrians jointly contribute to such an effect while avoiding behaviors between pedestrians play little role. The findings complement the existing empirical observations and lay a foundation for realistic pedestrian modeling in retail areas.

Danrui Li
A Psychological Approach to Understanding Microscopic and Macroscopic Structures During Train Boarding Processes

Current research on the train boarding process focuses predominantly on the physical modelling of pedestrian dynamics and situational characteristics such as platform design. Little psychology is involved in this approach even though individual behavior is a major factor in influencing dwell times. We take a systematic psychological approach to estimate whether parameters like pedes-train speed, area available per pedestrian, and interpersonal distance show variation at the microscopic level (individual variation), macroscopic level (situational variation), or both. Analyzing real-life train (de)boarding events (n = 3728) at a specific location at a Dutch train station, we find that boarders’ speed varies more at the microscopic (individual variation) than the macroscopic level. This behavior could thus result from stable aspects of the situation such as some structural feature of the environment or a social norm. Understanding such variation is help-ful in designing behavioral interventions/nudges. If variation is due to individual differences, then an individual-targeted intervention will be most effective. If variation is due to situational differences, then individual-targeted interventions may not be particularly useful, and nudges targeted at crowds or environmental features may be most effective.

Rabia I. Kodapanakkal, Caspar A. S. Pouw, Gunter Bombaerts, Alessandro Corbetta, Andrej Dameski, Antal Haans, Jaap Ham, Andreas Spahn, Federico Toschi
Empirical Comparison of Different Pedestrian Trajectory Prediction Methods at High Densities

Predicting human trajectories is a challenging task due to the complexity of pedestrian behavior, which is influenced by external factors such as the scene’s topology and interactions with other pedestrians. A special challenge arises from the dependence of the behaviour on the density of the scene. In the literature, deep learning algorithms show the best performance in predicting pedestrian trajectories, but so far just for situations with low-densities. In this study, we aim to investigate the suitability of these algorithms for high-density scenarios by evaluating them using two error metrics and comparing their accuracy to that of knowledge-based models. The first metric is distance-based, while the second counts the number of collisions between pedestrians. Our findings reveal that deep learning algorithms provide improved trajectory accuracy in the distance metric, but knowledge-based models perform better in avoiding collisions.

Raphael Korbmacher, Huu-Tu Dang, Antoine Tordeux, Benoit Gaudou, Nicolas Verstaevel
Reusable Software Structures for Coupling Agent-Based Locomotion Models and Disease Transmission Models

The COVID-19 pandemic sparked interest in coupling locomotion and disease transmission models because numerical experiments conducted with these combined models can help understand the transmission process. To review and reproduce simulations or to transfer them to a broader context, the underlying simulation program must be accessible and engineered carefully. Current software for agent based infection models often does not satisfy these quality requirements. Therefore, we address the research question of how to build reusable and quality-assured software that simulates disease transmission in moving crowds. We examine this by integrating an infection model into the crowd simulation program Vadere. Our methods are rooted in computer science. The software project is open-source, which allows for retracing its development and external contributions. Vadere’s new feature enables users to add and adapt modeling approaches and to compare them.

Simon Rahn, Gerta Köster

Granular and Active Matter

A Closed Network of RNA Polymerase Flow Models for Analyzing Intracellular Transport

We present a network of several RNA polymerase flow models (RPFM) interconnected through a finite pool of resources. We prove that the network always approaches a steady-state by using tools from the theory of cooperative systems having a first integral. Through theoretical framework and simulations, our analysis shows that increasing any of the forward (backward) rates in any of the RPFMs yields a local effect, an increase (decrease) in the output rate of this RPFM, and a global effect, the output rate of other RPFMs all increases or all decreases. Through simulation, we also show that sometimes increase in backward rates in an RPFM is related to an increase in the total output rate of the network. We believe that the network of RPFMs can provide deep insights into analyzing many natural and artificial systems.

Aditi Jain, Arvind Kumar Gupta
Modified Version of Open TASEP with Dynamic Defects

We propose a modification to the study of site-wise dynamically disordered totally asymmetric simple exclusion process (TASEP). Motivated by the process of gene transcription, a study in ref. [39] introduced an extension of TASEP, where the defects (or obstacles) bind/un-bind dynamically to the sites of the lattice and the hopping of the particles on lattice faces a hindrance if the arrival site is occupied by an obstacle. In addition, the particle is only allowed to enter the lattice provided the first site is defect-free. In our study, we propose that the particle movement at the entry of the lattice must face an equal hindrance that is provided by the obstacles to the rest of the particles on the lattice. For open boundaries, the continuum mean-field equations are derived and solved numerically to obtain steady-state phase diagrams and density profiles. The presence of obstacles produces a shift in the phase boundaries obtained but the same three phases as obtained for the standard TASEP. Contrary to the model introduced in ref. [23], the idea to introduce the modification at the entrance shows that the limiting case $$p_{d} \to 1$$ p d → 1 converges to the standard TASEP, where $$p_{d}$$ p d refers to the affected hopping rate due to presence of obstacle. The mean-field solutions are validated using extensive Monte Carlo simulations.

Nikhil Bhatia, Arvind K. Gupta

Cities, Vehicular Traffic and Other Transportation Systems

Oscillation Growth in Mixed Traffic Flow of Human Driven Vehicles and Automated Vehicles: Experimental Study and Simulation

This paper reports an experimental study on oscillation growth in mixed traffic flow of automated vehicles (AVs) and human vehicles (HVs). The leading vehicle moves with constant speed in the experiment. The following vehicles consist of six programmable AVs and different number of HVs. Thus, the market penetration rate (MPR) of AVs decreases with the increase of platoon size. The constant time gap car-following policy is adopted for the AVs and the gap is set to 1.5 s. The experiment shows that in the 7-vehicle-platoon, the oscillations grow only slightly. In the 10-vehicle-platoon, the AVs could still significantly suppress the growth of oscillations. With the further decrease of MPR of AVs in the 13- and 20-vehicle-platoon, the AVs become having no significant impact on oscillation growth. On the other hand, the flow rate of mixed platoons is lower than that of All-HV platoons of the same size. A simulation study is carried out, which exhibits good agreement with the experiment.

Shiteng Zheng, Rui Jiang, H. M. Zhang, Junfang Tian, Ruidong Yan, Bin Jia, Ziyou Gao
Modelling the Influence of Amber Light Dilemma Zone on Driver Behaviour Under Mixed Traffic Conditions

Signalized intersections are considered as one of the safest type of intersections, however a number of crashes are occurring in these intersections. One of the major causes of these fatal crashes is the presence of amber light di-lemma zone. The main motive of the study is to find out the different factors which influence the driver’s decision making under the influence of amber light dilemma. The study is carried out in Thrissur district of Kerala. Video data was collected from three different intersections for a duration of three hours each-Binary logit model for all vehicle types were prepared in order to determine how the heterogeneous traffic conditions influence the driver behaviour under the influence of amber light dilemma. A combined vehicle model considering all the vehicle types was also developed. The variables which were found to be significantly affecting the driver’s decision making were distance to stop line, approach speed, crossing distance and amber time. Among different vehicle types two-wheelers were found to have a significant influence on driver behaviour. Using the intersection details collected and the vehicular and geometric characteristics obtained from previous approaches the safe crossing distance and safe stopping distance of different vehicle types for different speed ranges were calculated. The insights from this study can be used to enhance the safety and performance of signalized intersections in developing countries.

K. R. Abhijith, K. Krishnamurthy
Optimal Design of Battery, Charging Infrastructure Planning, and Charging Scheduling for Electric Bus Network

Recognizing the growing global threat of air pollution, transit agencies and governments across the world are focusing on switching to electric vehicles (EVs). Battery electric buses (BEB) are becoming a crucial component to plan a sustainable transportation system. Two biggest drawbacks of BEBs are range anxiety and higher charging time. To overcome these hurdles, decision-makers are required to execute strategic decisions such as planning and design of charging locations as well as operational decisions like scheduling the charging activities for BEBs. This study gives insight into the planning and design of charging infrastructure network for BEBs while minimizing the total annual cost of designing the BEB system Apart from this, this study also focuses on the scheduling part for BEBs and provide information on the occurrence of charging activities for BEBs. A mixed-integer linear programming model is formulated which models the trade-off between route-specific battery capacity and terminal-level charging power (charger size along with the number of chargers at a terminal) and generates a scheduling plan for charging the BEBs. A part of New Delhi’s public bus network identified for electrification (18 bus routes and 21 terminal stations) by its transit agency was selected for our model application. Results suggested that model was able to obtain trade-off between battery size allocation and charger size selection.

Tanisha Pangtey, Pranav Gairola, N. Nezamuddin
The Impact of Vehicular Heterogeneity on the Rear-end Crash Risk in Mixed Traffic: An Extreme Value Approach

Traffic conflicts are most used surrogate safety measure (SSM), primarily defined using temporal or spatial proximity indicators assuming homogeneous traffic scenario where vehicle sizes do not vary much. In order to segregate conflicts from normal interactions, a proper threshold value of a conflict indicator must be used. Selecting an appropriate threshold value is challenging since it de-pends on static and dynamic characteristics such as size, speed, acceleration, and breaking capability of the vehicles. Inappropriately selecting the threshold would likely lead to false estimation of conflict and hence safety. The objective of pre-sent study is to define conflict considering vehicle sizes for more accurate rear-end crash risk estimation in heterogeneous traffic conditions. Time-to-collision (TTC), a conflict indicator was estimated from Traffic video data recorded at 4 unsignalized T-intersections for 3.5 hours at each location, identified as black spot on divided high-speed roads in Uttar Pradesh, India. In this study, a semi-automated method was used to extract vehicle trajectories from videos. In this paper, we propose extreme value-based Peak over threshold modelling to relate rear-end crash risk with vehicle size. Results show that threshold for TTC is dependent upon leader-follower pair. Therefore, conflict analysis based on global threshold would be biased since such normal interactions may be identified as critical even if they are safe. The proposed framework can be used for a more accurate risk assessment and calibration of vehicle warning system in heterogeneous traffic conditions.

Ashutosh Kumar, Abhisek Mudgal
A Comprehensive Review of Car-Following Models: Heterogeneous Non-lane-based Traffic Viewpoint

Car-Following Models (CFMs) are the mathematical representations of interaction between two sequential vehicles, leader and follower, in the same lane. Most studies on car-following have been conducted on lane-based traffic in developed countries. The mixed traffic of developing countries with high heter-ogeneity and weak to no lane discipline create a complex system which is quite challenging to model. There are quite a few commendable studies on modelling and understanding the complex driving behaviour of the mixed traffic. The pre-sent study aims to provide an organized review of appropriate longitudinal move-ment models/CFMs in the context of mixed traffic and reviews the studies on critical factors influencing longitudinal driving behaviour. Contemplating these determinants, this study investigates four CFMs. Further, the strengths and limi-tations of each model are discussed, and the possibilities of their combination, extension or modification are explained. These models consider two or more of the seven determinants shortlisted in this study. The importance of each determi-nant factor and its role in better understanding the microscopic behaviour of mixed traffic are also discussed.

H. R. Surya, Akhilesh Kumar Maurya, Shriniwas Arkatkar
Travel Path Tracking Using Smartphone Inertial Sensors: An Experimental Study on an Academic Campus Road Network

Travel path-related information plays a key role in understanding the route choice behavior of users in metropolitan cities. In this paper, the authors investigated the application of smartphone inertial sensors such as accelerometer, gyroscope and magnetometer for collecting the travelled path information of a trip maker. A framework was proposed to predict the actual travelled path that uses the smartphone's inertial sensor data collected from the trip makers and a detailed road network database. The proposed framework was tested on an academic campus road network with three different commuting vehicles. A total of 100 trips were collected using three different vehicles, namely, bus, car and Erickshaw, out of which 22 trips were used for training the decision tree algorithm. The algorithm is able to predict the path features with 95.6% accuracy. The effectiveness of the proposed framework was evaluated with 77 trips corresponding to E-rickshaw. The results show that the proposed framework was able to predict the actual travelled path with 79.22 % accuracy, and the travelled path could be predicted within the top ten alternative routes with 100% accuracy. Thus, the framework presented in this study shows a huge potential for predicting the actual travelled path without using the location sensor data.

V. A. Bharat Kumar Anna, Mallikarjuna Chunchu, Venkatesh Tamarapalli
The Intelligent Agent Model—A Fully Two-dimensional Microscopic Traffic Flow Model

Recently, a fully two-dimensional microscopic traffic flow model for lane free vehicular traffic flow has been proposed [Physica A, 509, pp. 1–11 (2018)]. In this contribution, we generalize this model to describe any kind of human-driven directed flow including lane-based vehicular flow, lane-free mixed traffic, bicycle traffic, and pedestrian flow. The proposed intelligent-agent model (IAM) has the same philosophy as the well-known social-force model (SFM) for pedestrians but the interaction and boundary forces are based on car-following models making this model suitable for higher speeds. Depending on the underlying car-following model, the IAM includes anticipation, response to relative velocities, and accident-free driving. When adding a suitable floor field, the IAM reverts to an integrated car following and lane-changing model with continuous lane changes. We simulate this model in several lane-based and lane-free environments in various geometries with and without obstacles. We observe that the model produces accident-free traffic flow reproducing the observed self-organisation phenomena.

Martin Treiber, Ankit Anil Chaudhari
Modelling Impact of Lateral Behaviour of Successive Vehicles on Traffic Safety for Regular and Work-Zone Roads

Highway movement is a reproving mode of transportation in the con-text of accessibility and mobility. Highway construction and maintenance activities are prevalent in India because of increased road transportation clamour. A work zone is a trapped portion of the road where construction and maintenance activities occur. These activities make up multiple types of work zones. Crashes and conflicts have increased in both regular and work-zone road sections in India, where road collisions increased by 10% in 2021 compared to 2020. The lateral behaviour of vehicles has a considerable effect on traffic safety. Hence, it is imperative to study the impact of the lateral behaviour of successive vehicles on traffic safety for regular and work-zone roads.

Omkar Bidkar, Shriniwas Arkatkar, Gaurag Joshi, Said M Easa
Stochastic Optimal Velocity Model with Two Vehicle Control Methods

Automated vehicles are widely considered as an important element of future transportation systems, and their adoption is expected to reduce traffic jams. In this paper, we propose a new mathematical model, namely, the controlled-SOV (C-SOV) model, to investigate the effects of controlled vehicles on traffic flow. Our Proposed model incorporates vehicle control in to the well-known stochastic optimal velocity (SOV) model. We propose two control. The gap-based control method regulates the velocity of a controlled vehicle to smooth its front and rear gaps, whereas the flow-based control method regulates the velocity of a controlled vehicle to smooth its front and rear flow. The results of our simulations indicate that the gap-based control, there are density regions where the flow rate is lower than that in the case where no control is applied. On the other hand, the flow-based control method consistently improves traffic flow.

Kayo Kinjo, Akiyasu Tomoeda
Machine Learning Approach for Modeling the Lateral Movement Decisions of Vehicles in Heterogeneous Traffic Conditions

The article describes the modeling of lateral movement decisions of motorized passenger vehicles like Cars, Motorised Three Wheelers(3W), and Motorised Two Wheelers(2W) under heterogeneous traffic conditions. The supervised machine learning approach was used to predict the lateral movement decision by treating the decision of lateral movement as a multi-class classification problem. Based on surrounding vehicles' information, a set of parameters was identified that potentially affect the decision-making process of drivers to change their lateral position. With the help of these parameters, the prediction ability of machine learning algorithms was compared. It was identified that these algorithms could predict the lateral movement decision of vehicles with an acceptable accuracy range. The results revealed that the random forest method out-performed all other algorithms and appeared to be a potential contender for modeling lateral movement decisions. Real-time position information about nearby vehicles may be gathered using advanced sensors and analyzed using developed models, allowing for the provision of safety features linked to lateral movement.

Prashant Baviskar, Shriniwas Arkatkar, Anshuman Sharma
Exploratory Data Analysis of Lateral Clearance Between Vehicles at Signalized Intersection with Weak Lane Discipline

Mixed traffic and weak lane-based driving habits of drivers in developing countries like India pose a significant challenge to a precise analysis of traffic condition and behavior. For weak lane disciplined mixed traffic conditions observed in India, the lateral driving behavior parameters are pivotal in assessing and analyzing the traffic proper effective monitoring, management and opera-tions. This study focused on exploring the lateral safety spacing maintained by vehicles observed at signalized intersections. The results show that the lateral clearance maintained by different vehicle classes while interacting with the vehicle travelling abreast to the subject vehicle is statistically different. The lateral clearance maintained by vehicle classes on the left and right sides is also significantly different. However, similar lateral clearance values can be considered for modelling for similar vehicle class combinations, but different values need to be considered for interactions between different vehicle classes.

Ritvik Chauhan, Ashish Dhamaniya, Shriniwas Arkatkar
Two-Dimensional Following Behavior Analysis of Powered Two-Wheelers Using Copula Approach

This paper reports a comprehensive investigation regarding the staggered following behavior of the powered two-wheelers (PTWs) in multi-class disordered traffic. The real-world trajectory data of such traffic stream was collected for this purpose and processed further to obtain the information on essential characterizing variables. The study adopts a two-layer framework. In the first phase, the spatial variables that characterize such behavior were identified, and their variation pattern was observed. More importantly, answers to two specific questions regarding the movement pattern of PTWs was obtained: (i) Whether the movement pattern of PTWs differ according to the type of lead vehicle they are following? (ii) Does it change according to the side in which the following is done? The second phase aimed at modeling the dependence between lateral and longitudinal descriptors of PTWs following behavior to precisely describe its movement pat-tern. The applicability of the copula approach was investigated for this purpose, and the obtained results indicated that Frank copula could model the joint distribution for almost all leader-follower vehicle pairs. The results also highlighted the importance of considering lateral separation for a realistic representation of multi-class disordered traffic streams. The developed models would contribute to better simulation models to represent the movement of PTWs in such traffic streams.

Rushikesh Amrutsamanvar, Lelitha Devi Vanajakshi
Analysis of Traffic Jerk Effect in a New Lattice Model with Density-Dependent Passing

In real traffic dynamics, non-motor vehicles often undergo sudden acceleration changes while passing, leading to traffic congestion on roads. Thus, a lattice model is modified to examine the impact of traffic jerk, taking into account the density-dependent passing behavior. The linear stability analysis is performed. It is found that the stability region reduces considerably with an increasing traffic jerk coefficient. By the reduction perturbation method, the kink-antikink soliton wave solution of the mKdV equation is attained, which describes the propagation of the density wave near the critical point. The theoretical results are validated by numerical simulation.

Muskan Verma, Sapna Sharma
Analyzing the Operational Performance of Mixed Traffic Comprising Autonomous and Human-Driven Vehicles at Varying Penetration Rates on Indian Urban Arterials

The autonomous vehicle (AV) market is in a nascent stage; as time progresses, more users will shift to AVs. This shift from conventional/human-driven vehicles (HDV) to AVs will have an impact on traffic systems in the future. Initially, the percentage of AVs will be less on the roads, but as the time and market penetration rate (MPR) progress, there could be a higher percentage of AVs on roads. Urban arterials in the Indian context are characterized by several traffic issues such as congestion, non-lane-based traffic, unsafe driving practices, etc. The traffic congestion in such urban areas will only increase going forward as the country develops, which reduces the operational performance of such traffic facilities. The shift from HDV-only traffic to mixed traffic consisting of AVs and HDVs will influence the operational performance of the arterials. The effect on operational performance due to increasing AV percentage on roads needs to be explored further to understand the implications of AVs on HDVs. Further, AVs have interesting properties such as better assertive movements than HDVs, and a higher ability to understand and comprehend complex traffic situations. This study attempts to understand the effect of increasing the MPR of AVs on the operational performance of traffic systems. A micro-simulation tool (VISSIM) is used to quantify the effects of MPR on the operational performance of urban arterials. Results show that as the market penetration of AVs increases the performance of the urban arterials improved in terms of enhanced capacity, space utilization and other microscopic characteristics.

K. N. Krishnan, K. Ramachandra Rao
A Jam-Absorption Driving System Based on Moving Jam Propagation Speed Estimation with Camera Sensors

In this study, a real-time control system is developed to operate the novel jam-absorption driving (JAD) strategy against single moving jams on freeway sections. Mixed traffic (human-driven vehicles and connected and automated vehicles (CAVs)) and heterogenous traffic (cars and trucks) are considered. The system calculates the moving jam propagation speed using vehicular trajectory data collected by camera sensors covering the entire control zone. The central controller selects a suitable CAV as an absorbing car in traffic through vehicle-to-infrastructure communication for performing JAD. Simulation results show that the system effectively enhanced traffic safety under a low market penetration rate of CAVs (approximately 1% to 10%). The investigation also shows the robustness of the proposed JAD system under various inflow rates.

Siyu Li, Ryosuke Nishi, Daichi Yanagisawa, Katsuhiro Nishinari
Fuzzy Logic Based Automation of the Extraction of Surrogate Safety Measures and the Creation of Severity Classification Using Video Data

This paper uses fuzzy logic to create an Artificial lntelligence (A.I.) based automated system that mimics expert rating of traffic conflicts. Video data from multiple sites were collected to study traffic conflicts as surrogate safety measures for traffic collisions. As part of that effort human trained subjects were given instructions to analyze traffic conflicts and assign severity levels to those. This paper proposes a fuzzy logic-based A.I. system that is trained based on such data so that the process of severity assignment can be done by the software system.

Pushkin Kachroo, Anamika Yadav, Md Mahmudul Islam, Ankit Kathuria, Shaurya Agarwal
A Graphical Tool for Planning and Real-Time Operation of Freight Trains

Indian railways are one of the four largest railway systems globally and transport more than a billion tonnes of freight each year. However, several of the routes share the tracks with passenger trains. Passenger and freight trains operate at distinct speeds, and a slow-running freight train is often stopped to pass a faster passenger train. This leads to a reduction in the average speed of the freight train. In this context, a pre-processing tool to assist in the operational planning of freight trains using a trajectory-based graphical approach was developed to insert feasible rail movements in real time. The proposed method and its effectiveness are demonstrated using real-world data, which shows that the proposed method can greatly reduce delays and provide significant benefits for real-time applications.

Abhishek Raj, U. Sethu Vinayagam, Bhargava Rama Chilukuri
Approaches for Modelling Travel Time Uncertainty

Recent research outcomes have reasonably stated that network traffic is stochastic, hence traffic demand management policies like perimeter control and gating should embrace stochasticity in any form for efficiency. This work attempts to model Travel Time Distribution in terms of Travel Time Uncertainty (TTU), a stochastic quantity. The study advised splitting the traffic network into an appropriate number of subnetworks using a clustering technique to provide approximate spatial homogeneity since TTU is also contributed by spatial variability. Travel time in distinct time stamps is used to represent TTU initially since it is directly related to TTU, however the concerned models have proven deterministic behavior approximately. The study modelled the quantity using the Bureau of Public Road Link-function (BPR) with TTU and related factors. The redesigned model captured stochastic behavior and integrated heterogeneity nicely. The Modified BPR (MBPR) function addresses the hysteresis phenomena in trip time distribution versus traffic flow values, which few research have reported. TTU and deterministic descriptors allow for network-level stochastic research in this paradigm.

Shubham Parashar, Ninad Gore, Shriniwas Arkatkar, Said Easa
An investigation of traffic speed distributions for uninterrupted flow at blackspot locations in a mixed traffic environment

Modelling traffic characteristics is the foundation for resolving various traffic and transportation issues. Among them, traffic speed has a significant impact on roadway crashes at blackspot (BS) locations. Speed is a random variable; several studies have recommended normal distribution to characterize the distribution of traffic speed for uninterrupted flow. However, a mixedtraffic situation causes heterogeneity, and the distribution of speeds deviates from the normal distribution. The present study investigates the distributions of traffic speeds for uninterrupted flow at 18 blackspot locations and individual vehicle types in mixed-traffic environments. Seven distribution models, namely Normal, Lognormal, Gamma, Logistic, Weibull, Burr, and Generalized Extreme Value (GEV), are considered to determine the speed characteristics. Different parametric distribution models are fitted to the vehicular speeds using maximum likelihood estimation (MLE) methods. Kolmogorov-Smirnov (KS), Anderson-Darling (AD), and two penalized criteria, i.e., Akaike and Bayesian Information Criteria (AIC and BIC), are used as goodness-of-fit (GoF) measures to find the best-fitting distribution. The overall suitability of each predicted distribution is also determined using a novel ranking method. The test findings suggest that GEV and Burr are the most suitable empirical speed distributions, with GEV fitting best above 96%. When the heavy vehicle composition (truck, bus, and tractor) is below 10%, 10–14%, 15–20%, and above 20%, it follows the Weibull, Gamma, GEV, and Burr distributions, respectively, in a mixed traffic environment.

Debashis Ray Sarkar, Parveen Kumar, K. Ramachandra Rao, Niladri Chatterjee, Sourabh B Paul
Data-Driven Prediction for Red-Light-Running at a TJunction

Running red lights is a serious road safety problem worldwide, which often leads to severe injuries and fatalities. Most recent works focus on identifying red-light-running behavior through surveillance cameras for punishment of violations. A few works predict the red-light-running behavior of drivers at intersections with Support Vector Machines (SVM) method. But they pay little attention to non-motor vehicles and the accuracy needs to be further improved. To address this problem, we conduct an observational experiment and construct a trajectory dataset (RedRun dataset) with the software Petrack. We also propose an Environment-Aware Red-light-running and Trajectory prediction Network (EA-RTN). It predicts the trajectories and red-light-running behavior of individuals (i.e. pedestrians, bicycles, electric vehicles, tricycles and cars) at T-junctions to help road users judge others’ movement in advance. Specifically, EA-RTN consists of two modules: one is a fully connected neural network (FCNet), which uses two hidden layers to predict whether a road user will run a red light. The other is a two-layer long short-term memory neural network. It predicts the trajectories of road users in the next 2 seconds and then assists drivers to plan ahead. The losses of these two tasks are combined to update the weights for realizing the multi-task learning. To evaluate our model, experiments are conducted on RedRun dataset. The results show that our approach predicts red-light-running behavior of road users more accurately. The accuracy is about 10% higher than SVM method.

Sainan Zhang, Jun Zhang, Weiguo Song, Longnan Yang
Prototype Models for Predicting Vehicle Types Generated in Heterogeneous Traffic Simulation

Traffic in emerging countries often comprises various types of vehicles, i.e., it is heterogeneous. Previous studies have proven that the order of the vehicle types affects the properties of traffic, and the present authors confirmed that grouping behaviors exist in the order of field traffic. It is possible to incorrectly evaluate traffic if one randomly generates vehicles in a traffic simulation. In this study, to accurately replicate such spatial patterns, we compared the performances of prototype models for prediction the types of vehicles generated in a simulation area. Prototype models, including Gaussian-process (GP) generators and evidential deep learning (EDL) generators, increase prediction performance. However, different behaviors for random and patterned sections in traffic have also been observed. This study contributes not only to developing accurate traffic simulation models, but also to understanding the internal emerging structures of heterogeneous traffic.

Akihito Nagahama, Takahiro Wada, Keiki Takadama, Daichi Yanagisawa, Katsuhiro Nishinari, Kenji Tanaka
Non-poissonian Cellular Automaton Models for Vehicular Traffic

In this study, the non-Poissonian (NP) versions of four types of cellular automaton models for vehicular traffic, namely, the totally asymmetric simple exclusion process (TASEP), stochastic Fukui-Ishibashi (SFI) model, quick-start (QS) model, and slow-to-start (S2S) model, have been investigated. In these NP models, the standby time for the next movement of each particle follows an arbitrary probability distribution, therefore, the disorder of each particle movement can be controlled by the coefficient of variance (CV). Fundamental diagrams show that the flow increases when the CV decreases in all four models. A substantial improvement in the flow against the TASEP (without any extension) is observed when the CV is large in the SFI and QS models. However, such a significant impact is seen when the CV is small in the S2S model.

Daichi Yanagisawa, Takahiro Ezaki, Akiyasu Tomoeda, Katsuhiro Nishinari
Estimation of Travel Time Reliability Using Wi-Fi Detections on an Urban Arterial Road

This study examines the variations in travel time and travel time reliability on an urban arterial route in Chennai, India, by using Wi-Fi detections. The travel time data was collected over a seven-week period using three Wi-Fi sensors. The study used Generalized Extreme Value (GEV) distribution to analyze travel time variations and found it to be the best fit for the data. It was found that the time of day (TOD), day of the week (DOW), and direction of travel (DOT) have a significant impact on travel time variations. The consistency of these variations is better measured through travel time reliability (TTR) measures. Disaggregate level comparisons of selected TTR measures were carried out to analyze TTR by TOD, DOW and DOT. For a given TOD, weekends are more reliable compared to weekdays. Similarly off-peak hours for a given a DOW were reliable than peak and off-peak hours. Additionally, TTR measures also revealed significant distinctions between segments, which can be attributed to varying traffic flow conditions.

Vikram Singh, Ninad Gore, Shriniwas Arkatkar
Dimensionality Reduction and Machine Learning-Based Crash Severity Prediction Using Surrogate Safety Measures

This paper presents a mechanism for using Machine Learning (ML) methods to extract the information efficiently from the input data collected in the field for predicting crash severity predictions. The output from the EVT engine subsequently can be used for approximately predicting traffic crashes. In our study, we use Principal Component Analysis (PCA) and Multidimensional Scaling techniques to obtain reduced dimensionality of the information obtained from multiple variables. Logisitic and Poisson regressions are used on the reduced number of variables which are the principal components, for crash severity predictions.

Pushkin Kachroo, Anamika Yadav, Ankit Kathuria, Shaurya Agarwal, Md Mahmudul Islam
Mean Field Games Modeling for Dynamic Traffic Assignment with Information

Dynamic traffic assignment describes how the demand of travellers spreads in time (departure time choice) and space (route choice). Information technology for traffic management is progressing fast, thus the information provided to travellers by various agencies has a strong impact on their dynamic assignment. The object of the paper is to analyze the impact of traffic information at the network level in the perspective of mean field game theory. In the proposed approach, route choice results from real-time information which provides travellers with instantaneous travel times (ITTs). It is assumed travellers chose the shortest path to destination with respect to ITTs. The day-to-day departure time choice is based on user equilibrium with each traveller minimizing his generalized origin-destination travel cost. The travel cost includes the effective predictive travel time and the late/early arrival time penalty. The user departure time equilibrium is formulated as a mean field game, a paradigm which combines individual competition for ressources with global dynamics. A numerical example illustrates the convergence towards equilibrium of the proposed approach.

Megan M. Khoshyaran, Jean-Patrick Lebacque
Development and Field Validation of Nonlocal Velocity based Macroscopic Traffic Model

This paper presents a new nonlocal calculus-based macroscopic traffic model that eliminates the specific boundedness limitation that the previous local models had. The paper develops the model and demonstrates the analysis performed on the model. It then demonstrates a numerical model for its approximation and also a field validation using NGSIM traffic data. Results indicate its superiority to local models.

Pushkin Kachroo, Shaurya Agarwal, Animesh Biswas, Jiheng Huang
Connected and Autonomous Vehicle’s Behavior in Heterogenous Disordered Traffic in Metropolitan Cities

The government of India introduced the smart city program in 2015 to make the selected city smart and provide all the necessary services required for the public. The smart city program will become much more efficient with the introduction of artificial intelligence, information and communication technology, and sensors technology. These technologies facilitate the necessary infrastructure for the deployment of connected and autonomous vehicles (CAVs) in these cities of India. Self-driving technology has only recently been introduced in commercial vehicles however, the integration of these vehicles into a CAV ecosystem is still being tested. In a study conducted by the Victoria Transport Policy Institute in 2015, it was estimated that by 2040, 75% of all vehicles will be autonomous, hence we could soon see the implementation of Autonomous Vehicles and its integration into a network in India. However, the deployment of CAVs could be faced with several immediate challenges with respect to road users: from Vulnerable Road Users (VRUs) to other motorized vehicles. Studies on the challenges of CAV deployment have been conducted in several developing economies, but there exists a gap when it comes to understanding the potential challenges that could occur in an Indian context. Hence, this paper focuses on analyzing the behavior of CAVs in a heterogenous disordered traffic such as in Indian metropolitan cities. This analysis is based on a systematic review of existing literature and policies. This review focuses on specific aspects based on the Indian scenario such as the sensing capabilities required for efficient functioning of CAVs, driving behavior related requirements, contextual challenges of road and traffic environments and regulatory challenges. The outcome of this study is the identification of these challenges in the deployment of CAVs in Indian metropolitan cities and the formulation of possible solutions for the identified problems.

Aditya Verma, Suresh Chavhan
Synergy of Model-driven and Data-driven Approaches in a Dynamic Network Loading Problem

Modern dynamic models of traffic flow and especially dynamic network loading (DNL) models are a powerful approach to predict traffic flow dynamics in a short-term sense (minutes or hours ahead). Such models should be the core element of any intelligent transportation system to make safer and smarter use of transport networks. Nowadays a variety of traffic data is becoming more and more accurate and available. Online traffic data can be incorporated in DNL model to take into account nonrecurring events (e.g. accidents, road closures or unexpected bad weather conditions). This idea can increase the accuracy of short-term prediction and make traffic flow management more effective. In our research we suggest to combine traditional model-driven approach with a data-driven prediction. As a DNL model we use the link transmission model in cooperation with a dynamic user equilibrium algorithm to identify the routes. Traffic data are the values of speed and flow with a 5-minutes time step, obtained from stationary road sensors. We use the rolling horizon approach, that is, every 5-minutes model constructs 1-hour forecast incorporating actual sensor data. Moreover, we use methods of machine learning to predict the sensor data for the next hour and take it into account while calcu- lating the forecast for the current hour ahead.

Valentina Kurtc, Andrey Prokhorov
Sensor Data Analysis by means of Clustering

Traffic jams are a big problem of the society nowadays, especially in case of urban traffic. To solve the problem of traffic congestion and air pollution, the intelligent transportation systems (ITS) should be developed and integrated into transport infrastructure. The core element of such ITS is a reliable and accurate forecasting model to predict traffic flow in a short-term period. Lots of historical traffic data can be used as input of the model, in particular daily traffic profiles. Different dates have different traffic flow patterns, and modern prediction models should take into account such temporal variations. This paper investigates the historical traffic flow data, which was obtained from stationary road sensors. The main goal of this research is to obtain more insight into urban traffic by analyzing between day and between month variations in traffic volumes. By means of k-means clustering procedure, we divide daily traffic profiles of each sensor into several groups and examine the obtained clusters with respect to day type (day-of-week, preholiday and holiday) and seasonal variations. For most road sensors, there is a significant difference between the daily traffic flow profiles for working and non-working days. The centroid of the first one demonstrates two prominent flow peaks (morning and evening), whilst the second one represents only one day peak with slow growth and slow decrease of flow rate. We discovered seasonal variation for some road sensors, but it is less pronounced than the variations between weekdays.

Valentina Kurtc, Andrey Prokhorov
Delay Modelling at Signalized Intersection Under Mixed Traffic Conditions

Delay is one of the major parameters in estimating the level of service (LOS) of a signalized intersection. Several methods are available for estimating the delay at a signalized intersection. Highway Capacity Manual (HCM) method and Webster’s method are the most popular and widely used methods among them. These methods are based on homogeneous traffic conditions. But in developing country like India, traffic condition is highly heterogeneous with almost no lane discipline. So, these conventional methods will give errors in de-lay estimation and a proper delay model is needed as per Indian traffic conditions. In the present study, a new delay model is developed by modifying Web-ster’s delay equation. Field delay was estimated according to HCM 2010 procedure. Variations of field delay with different input variables were studied. The semi-empirical adjustment term in Webster’s equation is replaced with the field delay adjustment factor. The adjustment term was developed with two methods, Artificial Neural Network (ANN) model and the regression method using SPSS software. Mean absolute percentage error (MAPE) and coefficient of determination (R2) were taken into consideration for determining a better result. Results obtained from both models were compared with the conventional method and the proposed model gives better results.

T. P. Akash, K. Krishnamurthy
Traffic and Granular Flow '22
K. Ramachandra Rao
Armin  Seyfried 
Andreas Schadschneider
Copyright Year
Springer Nature Singapore
Electronic ISBN
Print ISBN

Premium Partner