Table of Contents Author Guidelines Submit a Manuscript
Journal of Advanced Transportation
Volume 2018, Article ID 1063043, 42 pages
https://doi.org/10.1155/2018/1063043
Review Article

A State-of-the-Art Review on Empirical Data Collection for External Governed Pedestrians Complex Movement

1Jiangsu Key Laboratory of Urban ITS, Southeast University, China
2Jiangsu Province Collaborative Innovation Center of Modern Urban Traffic Technologies, China
3School of Transportation, Southeast University, 2 Dongnandaxue Rd, Nanjing, Jiangsu 211189, China
4School of Engineering, RMIT University, Carlton, Melbourne, VIC 3053, Australia
5Safe Transportation Research & Education Center, Institute of Transportation Studies, UC Berkeley, 2614 Dwight Way, Berkeley, CA 94720-7374, USA

Correspondence should be addressed to Zhirui Ye; nc.ude.ues@iurihzey

Received 24 April 2018; Revised 3 August 2018; Accepted 14 August 2018; Published 2 September 2018

Academic Editor: Shamsunnahar Yasmin

Copyright © 2018 Xiaomeng Shi et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Complex movement patterns of pedestrian traffic, ranging from unidirectional to multidirectional flows, are frequently observed in major public infrastructure such as transport hubs. These multidirectional movements can result in increased number of conflicts, thereby influencing the mobility and safety of pedestrian facilities. Therefore, empirical data collection on pedestrians’ complex movement has been on the rise in the past two decades. Although there are several reviews of mathematical simulation models for pedestrian traffic in the existing literature, a detailed review examining the challenges and opportunities on empirical studies on the pedestrians complex movements is limited in the literature. The overall aim of this study is to present a systematic review on the empirical data collection for uni- and multidirectional crowd complex movements. We first categorized the complex movements of pedestrian crowd into two general categories, namely, external governed movements and internal driven movements based on the interactions with the infrastructure and among pedestrians, respectively. Further, considering the hierarchy of movement complexity, we decomposed the externally governed movements of pedestrian traffic into several unique movement patterns including straight line, turning, egress and ingress, opposing, weaving, merging, diverging, and random flows. Analysis of the literature showed that empirical data were highly rich in straight line and egress flow while medium rich in turning, merging, weaving, and opposing flows, but poor in ingress, diverging, and random flows. We put emphasis on the need for the future global collaborative efforts on data sharing for the complex crowd movements.

1. Introduction

Pedestrian traffic is an essential mode in a multimode and multilevel transportation system. In contrast with vehicular traffic, pedestrians tend to change their speed and direction more frequently resulting in complex interactions with other pedestrians, transport modes, and the traffic infrastructures. These interactions can be generally categorized into four types: Pedestrian to Infrastructure (P2I), Pedestrian to Pedestrian (P2P), Pedestrian to Motor Vehicle (P2V), and Pedestrian to Nonmotorized Transportation (P2N) [1]. In transport hubs (e.g., inside train station) or mass gathering (e.g., a concert), the interaction types for pedestrian traffic are typically limited to P2I and P2P. The simple maneuver could be pedestrians walking relatively straight in one direction at uniform speed (unidirection movement) to their destination. However, as the direction of movements and speed changes (e.g., multidirection movements), this can lead to complex interactions and competitiveness among pedestrians [2]. The movements of pedestrians to a large extent are governed by external factors such as infrastructure constraints (e.g., people at bottlenecks are forced to slow down as a result of space competitions) and maintaining proximity from other moving objects (e.g., a pedestrian has to change his/her walking direction for collision avoidance purpose). As a result, we observe pedestrian crowd movement ranging from unidirectional to multidirectional movement at major public infrastructure such as train stations, sports stadiums, and other outdoor or indoor public gatherings. These multidirection movements can result in increased number of conflicts resulting in delays and congestion [3]. In addition, the additional delays due to conflicts can increase egress time during emergency evacuation thereby impacting the efficiency of the evacuation process [4]. Further, the additional delays can cause impatience among escapee resulting in pushing behavior which can lead to trampling and stampede [5]. Recent studies have discovered that the small architectural modifications in an escape area can have large potential effects in terms of outflow and safety of individuals [3, 611]. Therefore, pedestrians crowd movements and interactions with the escape area or infrastructure is a subject of greater research interests among researchers working in the field of transport engineering, infrastructure engineering, and architecture. As such, these complex multidirection movements have been studied extensively in the past [2, 3, 9, 12, 13]. Both microscopic and macroscopic approaches have been proposed to understand the mechanism of the pedestrian complex movements [3].

Recent advancements in computer processing power have prompted the researchers to collect and analyze the empirical data to a finer grain for model calibration and validation. Therefore, empirical data collections through controlled laboratory experiments have been on the rise in the past decade [2]. Although there are several reviews of mathematical simulation models for pedestrians’ traffic in the existing literature, only few studies have reviewed the state of the art of empirical data for pedestrian traffic [2, 14]. Even those limited studies have examined the empirical data collection for pedestrian traffic from broader perspective rather than exploring the detailed debates, challenges, and opportunities for multidirection complex movements. To the authors’ knowledge, there is no detailed and systematic review on empirical data collection for pedestrians’ complex traffic movement in the literature.

Therefore, the overall aim of this paper is to present a systematic review on the empirical data collection for uni- and multidirection complex movements. Considering the length and scope of this paper, we restrict the review to only uni- and multidirection movements governed by external factor by considering only P2I and P2P interactions. Please note that pedestrian traffic can also be governed by internal interactions among pedestrians which can lead to some self-motivated movement behaviors such as overtaking [15], self-slowing [16], queuing [17], and grouping [18], as well as crowds’ self-organization phenomena such as ‘faster-is-slower’ effect [5], “lane formation” [9, 19], “zipper effects” [20, 21], “herding” [22], and ‘freezing-by-heating’ [23]. These emergent behaviors are driven by the individual’s self-consciousness or the internal interactions within the crowd. These movement types have been identified as the underlying mechanisms of several crowd disasters [2433]. For example, under severely congested situation, the individual body’s movements are largely restricted by the surrounding pedestrians and any secondary increase of flow could result in the forward and backward compression waves leading to stampede accidents [24, 27].

The exclusive review of this internally governed behavior is not within the scope of this review. However, we have occasionally mentioned these phenomena when examining externally governed movement patterns at instances where the internally governed movements also played an important role in the efficiency of pedestrians walking operations. Also, this review is not intended to review the mathematical models for pedestrian traffic or the results obtained from simulation models on complex movements. Interested readers on simulation models for pedestrian traffic can look into the review work by Duives et al. (2013) [34]. This is necessary to balance the breadth and the depth of the review article.

Our paper is structured as follows. Section 2 describes the methodology of this literature review paper. Section 3 gives the classification methods of the complex movement pattern. Section 4 presents the discussion based on the reviews. Finally in Section 5, the summary of this review article is presented.

2. Methodology

We used Boolean searches of databases to obtain literatures from electronic database including ISI Web of Science, Scopus, and Google Scholar. Other online social platforms such as Researchgate and Mendeley as well as well-known conference proceedings in this field such as the Pedestrian and Evacuation Dynamics (PED), Traffic and Granular Flows (TGF), and Transportation Research Board (TRB) were also used as the search database. We first checked the literature review papers related to pedestrian traffic topic, especially multidirection movements, published in recent years. The literature search was then further supplemented by relevant publications in the reference lists of those recent publications.

The time period for the review is based on the past two decades (from 1998 to 2018) as it is the time-frame where we witnessed a significant advancement made to the data collection and modeling of pedestrian traffic. Only papers published in English are considered in this review as they are widely accessible to the researchers. We only reviewed the articles published in journal, conference, and book or book chapters. Therefore data from the government reports, industry related journals, or other sources are not considered. If the reviewed conference papers were extended and published in journals afterwards, we considered only the journal paper as the primary source and vice versa.

Reference management software, Mendeley Desktop, was adopted for the collection of literature, organization of the document, and generation of the cited list. This tool significantly improved the efficiency of writing the review paper.

3. Complex Movements of Pedestrian Flow

Complex movements are the aggregates of individual and crowd collective motions. To decompose the collective movement into critical movement behaviors, many researchers have studied the crowd motion behaviors both from pedestrian traffic system and other biological systems [35, 36].

However, a generic classification of pedestrian crowd behaviors is still missing from the literature [34, 35]. To the authors’ knowledge, a review on the hierarchy of the crowd motion behaviors with regard to the movement complexity is missing in the literature.

Duives et al. (2013) proposed a classification of crowd motion cases based on the flow patterns [34]. They decomposed the crowd movements into eight motion based cases belonging to two categories: unidirectional flow (straight flow, rounding a corner, entering, and exiting) and multidirectional flow (bidirectional flow, 2-strips crossing flows, >2-strips crossing flows, and random crossing flows without focal point). They stated that the mix of these cases could be able to cover the entire range of complex movements. In addition, they also summarized six types of self-organization phenomena from the literature: lane formation, ‘stop-and-go’ waves, turbulent flows, herding, the zipper effect, and the ‘faster-is-slower’ effect. However, the focus of that paper is the assessment of crowd simulation models, rather than the empirical evidences under each crowd motion phenomenon.

Based on the classification of interrupted and uninterrupted flows in vehicular traffic [37] and considering the predominant driving forces of complex interactions, we classify complex movements into two categories:(i)Externally Governed Movement. Movements are strongly restricted/affected by external factors such as infrastructure constraints (e.g., merging pedestrian crowd flow at T-intersection [38]). Pedestrians have to change their desired walking manner due to the competition of space and time. Therefore, under this category, movements are aggregated into patterns which have strong relationships with the geometrics of architectural configurations. In addition, the interpersonal interactions are also predominately affected by the external environments but not the internal factors.(ii)Internally Driven Movement. Movements are spontaneously organized within the crowd in an observable manner under appropriate situations. Under this category, the formulations of crowd movement patterns are less influenced by external factors. And the interpersonal interactions are usually driven by pedestrian’s willingness to behave his/her own manner or follow others’ behaviors (e.g., lane formation within a pedestrian crowd [19]).

In this paper, we build our review based on the eight classifications as stated by Duives et al. (2013) and discussed earlier. Additionally, we have extended and considered other complex movements that have been investigated theoretically or empirically in the literature such as turning, weaving, merging, and diverging movements as shown in Figure 1.

Figure 1: Conceptual diagrams of external governed complex movements.

Figure 2 shows the classification of complex movements that shows the category of pedestrians’ movements under external governed and internal driven movements. The external governed movements include unidirectional and multidirectional flows while internal driven movements include individual and crowd motion behaviors and self-organized phenomena. A dotted line representing the interface between internal driven and external governed movements is also drawn demonstrating that there could be overlap between these two types of movements (e.g., a merging flow may display ‘faster-is-slower effect’ or “lane formation”). As stated earlier in the introduction section, the scope of this paper is to review the externally governed pedestrian crowd movement. Figure 2 displays the conceptual diagram of these flow patterns as shown in the Figure 1 with a stepwise hierarchy. Starting with a simple straight unidirection movement, the pedestrian crowd movement can get complex as we move towards opposing, weaving, merging, diverging, and random flows. Please note that random flow can be the combinations of the above movement patterns. The greatest challenge for studying random flow is the uncertainty of conflicts. Therefore, the prediction of walking directions and potential conflicts is at the heart of this category.

Figure 2: Classification of complex movements.

As shown in Figure 2, the crowd dynamics becomes complex as the complexity of movement increases. Additionally, the figure also shows the connections and overlapping among these complex interactions.

Externally governed movement usually exhibits observable flow patterns as it usually occurs at fixed locations. In this category, we have put the externally governed movements into eight subcategories: straight line, turning, egress and ingress, opposing, weaving, merging, diverging, and random flows. The first three flow patterns (straight line, turning, and egress and ingress) belong to unidirectional flow, as the desired flow directions are united. Meanwhile, the others are affiliated with multidirectional flow, as the desired flow directions are varied (e.g., opposing, weaving, merging, and diverging). Please note that multidirectional movement like crossing can be a part of opposing flow or weaving flow.

Internal driven movement is more complicated because it can be observed only under certain conditions such as location and flow status. By expanding the summary of individual and crowd motion cases, which are primarily internal driven as highlighted in [39], we developed a comprehensive classification of the internal driven movement as shown in Figure 2. We first listed the self-motivated individual motion behaviors including overtaking [15], evading [40], following [41], and self-slowing [16], as well as the crowd motion behaviors, i.e., queuing [17] and grouping [42]. Then, we categorized 10 types of self-organization phenomena including herding [43], bubble effect [44], lane formation [45], stripe formation [9], zipper effect [20], oscillatory flow [46], faster-is-slower [5], freezing-by-heating [23], stop-go waves [47], and turbulent flow [48].

The above classification method is not restricted to a type of facilities, but is extended to the operations of pedestrians flow on those facilities. In addition, it should be noted that crowd dynamics are affected by both internal and external factors simultaneously, the movements in one category are not absolutely independence with the other category. Under some circumstances, the governing mechanisms of flow might be transformed to another category. Therefore, the overlapping among those interactions occurs as shown in Figure 2. Also, please note that the logical flow for each subcategory may vary. For example, some empirical studies contain more than one crowd phenomena, such as both unidirectional and multidirectional flows. In such case we separated the key findings but introduced the common experiment setups in their first belonging category.

In the following subsections, we describe the empirical studies for each external governed complex movement. Finally, a summary table highlighting the contributions and recommendations from those studies are presented at the end.

3.1. Straight Line

Straight line walking, as the basic form of walking, refers to the kind of unidirectional movement (i.e., without any direction change). It has been used as a benchmark for pedestrian dynamics studies. Datasets for straight line walking can be seen in several well-recognized handbooks in transportation and fire engineering fields as well as many early studies on pedestrian traffic [4955]. Some early year’s empirical data were not well documented as of missing details due to the limitation of the traditional data collection techniques. In addition, large discrepancy existed in these studies for the insufficient influence factor consideration and uncontrollability. In this section, we highlight the recent empirical data collected through more advanced techniques [48, 5659].

As shown in Figure 1, straight line walking can be segmented into single-file and multilanes movements. In single-file movement, pedestrian has no lateral interaction with other pedestrians; i.e., the interactions are purely longitudinal. Meanwhile, multilanes straight line movement requires the consideration of the horizontal oscillations from the other walking streamlines.

Single-file movement experiments were carried out with participants from different cultural background. It is worth mentioning that despite that many experiment setups in straight line walking are loop corridors/passageways, most of them have not considered the turning movements at corners and restricted the measuring areas to the straight parts. Such oval-shaped loop corridors are also different from the settings of ring-shaped corridors where pedestrian kept turning during their movements [57, 60].

Seyfried et al. (2005) set up experiments in a loop corridor with up to 34 German participants [57]. Through manual video analysis, velocity-density relationships for single-file movement were determined and compared with the movement on 2D plane. They found little difference between single-file movement and 2D plane movement in terms of density range and concluded that the internal friction and other lateral interventions had little influence on density-speed relations. Later, Liu et al. (2009) replicated Seyfried et al. (2005)’s setup and performed a similar single-file movement experiment involving Chinese participants [61]. Through the trajectories derived from automatic image processing, velocity, density, amplitude, and frequency of lateral oscillation, time headway values and distributions were calculated. They also observed that the lateral oscillations change with velocity or density. As density increased, time headway amplified while spacing reduced. After comparing the results with Seyfried et al. (2005), they found that their results exhibited higher speed under same density.

In the same year, Chattaraj et al. (2009) also carried out the same experiments in India to compare the fundamental diagrams of straight line movement with those from Seyfried et al. (2005) in Germany [62]. They found significant cultural difference between the two groups. For example, German group tend to have more minimum personal space than Indian group while German group was more sensitive to the increase in density than Indian group. In addition, they also discovered that the length of corridor have no influence on the fundamental diagram. Apart from the effect of corridor length, boundary condition effects on fundamental diagrams were also empirically examined. Zhang et al. (2015) set up the similar corridors with Seyfried et al. (2015) with closed and open boundaries [63]. The influence boundary conditions on flow properties including the fundamental diagrams and density transition over time were investigated. They found that boundary condition has an effect on the flow properties and the effect was exceedingly high at jamming state. Under same density level, the velocities were lower under closed boundary condition. Cao et al. (2016) also set up the similar single-file experiments with 80 young students and 47 aged people in China [64]. It was discovered that jams tend to occur in mixed group due to the variation of maneuverability and self-adaptive abilities.

Recently, Seyfried’s team further continued their researches on single-file movement by considering the stepping behaviors of pedestrian locomotion. Adopting the similar setups, Wang et al. (2018) organized 196 participants to walk in the oval passageway at different densities [65]. Step length, step frequency, swaying amplitude, and step synchronization were measured from trajectories. They found that power functions had better goodness-of-fit than linear function in terms of describing the relations between step length and frequency and speed. In addition, the relations between swaying amplitude and speed could be segmented into two regimes. Likewise, Cao et al. (2018) further analyzed the experiment data in Cao et al. (2016) with the similar analysis of stepping behavior with Wang et al. (2018) [66]. They discovered additional findings on age effects on stepping behavior, e.g., young people tend to control the lateral movement better than the old people. They also fitted the relations between step frequency and speed and found a quadratic fit in three groups of experiments which were different than those observed with Wang et al. (2018). Moreover, they observed no significant effect of height and gender on the fundamental diagram.

In terms of multilanes movements, Daamen and Hoogendoorn (2003) performed a series of controlled experiments with 60-80 human participants [56]. Their experiments included straight line walking scenarios with different desired speed and density levels (Experiment 1-3). Flow characteristics such as density and speed distribution, travel time, and fundamental diagrams were calculated from the extracted video trajectories. Likewise, Helbing et al. (2005) organized about 100 college students to walk through a straight passageway bounded with desks and chairs [9]. Bottleneck was created by moving the desk to narrow the passageway along the walking line. Mean flow and time headways were counted from videos. Results were compared between the corridor setups with and without bottleneck and they found significant capacity drop in bottleneck situation under unidirectional cases.

Similarly, Asano et al. (2007) conducted crowd experiments in a campus with 94 students [12]. Flow demand (30, 94) and controllability (with/without instruct to walk) were considered in straight line scenario. Multilanes walking features with different desired speed level in each stream were also investigated. Trajectories were extracted from video and headway-speed relationship for straight flow was calculated for different cases. Likewise, Zhang et al. (2011) performed a series of large-scale experiments in well-designed corridors with up to 350 people [58]. Controlling the number of pedestrians, width of corridor, width of entrance, and exits, 28 runs of straight line experiments were performed. From the individual trajectories extracted using a software called PeTrack [67], the influences of density measurement methods and corridor width on fundamental diagrams were analyzed. It was found that, for density less than 3.5 m−2, measurements method had minor influence and the specific flow was independent of corridor width as the density level was not highly congested.

More recently, Seyfried’s team reported their new large-scale experiments results in Cao et al. (2017) [13]. About 2000 participants were involved in this series of experiments, including multilanes straight line movements in a corridor with a width of 5 m. Fundamental diagrams for different density measurements were presented and the estimated specific flow for unidirectional flow was 1.4 Ped/m/s at density level of 1.5 Ped/m2. Likewise, Sharifi et al. (2017) conducted a series of laboratory experiments in a circuit corridor that contained straight line passageway [68, 69]. Heterogeneity of crowd dynamics was investigated through the involvement of 189 normal people and 42 disabled people [70]. Trajectories of participants were extracted through automatic video processing and the macroscopic and microscopic crowd motion parameters were calculated. Capacities and level-of-service for straight line facilities with disabled people were further analyzed through the modeling of time headway [71].

Several field studies of straight line movement on different types of walking facilities were carried out [7274]. To calibrate and validate Legion’s simulation model, Berrou et al. (2007) collected 4762 valid samples of pedestrians’ walking speeds from 18 hours of video recorded by 23 cameras in New York City Transit Station. The researchers compared their results (mean speed 1.50m/s with standard deviation of 0.21m/s) with several published free speed statistics. Similarly, to derive pedestrian’s walking speed distribution on a long stairway, Kretz et al. (2008) carried out a field observation at a stadium in Germany [72]. 485 pedestrians were observed and divided into three categories according to the degree of influence by surrounding people. Results showed that the maximum horizontal walking speeds distribution exhibited 0.4m-0.5m/s depending on density level. They also compared the speed distribution with a short stairway case and found the speed in a long stairway could be the half of that on a short stairway. They also discovered the gap of lack of a universal scaling factor for the speed on stairs relying on the length of stairway. Regarding high density situation in multilanes straight flow, Zhang et al. (2013) collected a field data of dense crowd during a mass gathering event through the combined use of camera and active infrared counter [73]. Flow rate and velocity of the crowd were measured and the fundamental diagrams were painted and compared with previous studies. Three additional flow operational indices including flow rate/velocity transition, velocity field, and speed map were calculated. Results suggested that the capacity for the street was estimated between 1.73 and 1.98 /m/s under normal situation and the density should be controlled under 5 /m2.

Apart from human subjects, several researchers also investigated the straight line movement through animal experiments and examined their relevancy to pedestrian crowd movement. John et al. (2009) reported the collective dynamics of straight line movements of ants on trails creating a nature-like experiments [75]. They observed no overtaking maneuvers in ants traffic trails but noticed the ants platoons that could be forecasted by simple models. It was observed that the flow increased monotonically with the density and no jam branch could be observed. Sharp contrast was found in ants traffic as compared with vehicular traffic and other transport modes where overtaking and congestion are frequently observed. John et al. (2009)’s findings were observed under nonstressed situations. To examine the straight line movement of stressed ants, Wang and Song (2016) made a passageway (1cm wide and 20 cm long) to measure the stressed ants’ movements with different body size [76]. Fundamental diagrams were presented for both small and large ants. They found no evidence of jamming in stressed ants trails which was in line with John et al. (2009)’s findings. Also, they observed that speed seemed to be constant with density. Likewise, they also noted the differences of ants traffic behavior comparing to vehicular and pedestrian traffic in terms of fundamental diagrams.

3.2. Turning

Turning movement can be regarded as the response to the requirement of changing walking direction. It is usually associated with walking maneuver around corners or at exits, angled corridors, or ring corridor scenarios [77]. For single turning movement, the number of walking streamlines does not change before and after turning process and turning flow does not interact with flows in other direction. Therefore, turning movement can be also categorized into unidirectional flow.

It is worth mentioning that turning movement in this review has different meaning than those gait-level of works such as body-turning behavior as the turning movement should show direction change in the general walking streamlines [78].

Literatures documented several simulation studies on turning movements [7982]. In addition, empirical evidences could be found with both human subjects and biological entities [81, 8385].

Regarding turning movements around the corners and exits, Yanagisawa et al. (2009) organized 18 participants to perform an experiment with a special attention on the turning locomotion at bottlenecks [83]. They found the theoretical relations of turning angle on the parameters of floor field model under different cases. They compared the theoretical results with empirical data and found the complete agreements. In addition, obstacle effects were also examined by placing a column with 20 cm diameter in front of the exit and the increment of outflow from the obstacle was observed. Apart from the straight line experiments, Zhang et al. (2012) also carried out turning experiments at the cornered corridor with 2.4m width [84]. Fundamental diagram of turning flows at a corner was observed. With regard to turning movements around stair corners, Burghardt et al. (2013) organized experiments to observe the turning movements on stairways [86]. They presented the fundamental diagrams and conducted profiles analysis of density, velocity, and specific flow at stairways corner and compared their results with four planning handbooks. They found that the turning bends could be the potential constraints for safety issue so that stairs without bend should be preferred by designers.

In terms of turning movements at angled-corridor, Dias et al. (2014) performed controlled experiments to obtain the microscopic characteristics of 16 individuals walkers in angle-changing corridors (0°, 45°, 60°, 90°, 135°, and 180°) under three different desired speed levels (normal walking, faster walking, and slow running) [77, 87]. They found the speed drop at a fixed turning region and the dimensions of such region was independent with turning angle but was dependent with desired speed levels. It was observed that the speed around the corner dropped with the increment of turning angle. Later, Dias et al. (2015) extended their previous turning experiments with up to 55 participants [88]. It was found that the longitudinal spacing between pedestrians tends to increase with the increased speed. The collected data were further utilized to develop an optimal trajectories algorithm and to calibrate a cellular automation model [79, 89]. Similarly, Sharifi et al. (2014) and Sharifi et al. (2017)’s experiments also included two angled corridors (right and oblique angle) as well as a spiral stairs with right corners [68, 90]. Walking speed distributions and time-space diagrams for both homogenous and heterogeneous flows were analyzed from the calculation of extracted trajectories. Their results exhibited a correspondence with Dias et al. (2014) as the speed reduction increased with the rise of turning angle [87]. Recently, Seyfried’s team reported the results of two sets of experiments in Sieben et al. (2017) [91]. One experiment set a right-angle corridor with two narrow entrances and more than 270 people were asked to enter the corridor. Density bursts and fluctuations were observed at turning sections. Apart from turning angle, the curve radius of turning region was also considered in a recent experiment in China. Sun et al. (2017) studied the combined effects of turning angle and bend radius on the operational characteristics of pedestrians [92]. Results indicated that, under different flow volumes, the cumulative density and average speed of pedestrian flow were affected by both angle and radius effect. Angle’s effect was more significant than radius, and the influence was maximized in right-angle case.

In terms of ring-shaped corridor scenarios, Jelić et al. (2012) performed single-file walking experiments with up to 28 pedestrians in a ring corridor with inner radius of 2 m and outer radius of 4.5m [93]. They derived their fundamental diagrams and instantaneous velocity and distance headway from both global and local measurements and compared the density-velocity relations with Seyfried et al. (2005) [57]. Their results showed that three linear regimes could be discovered from the velocity-spatial-headway relations which could be useful for further modeling. Likewise, Moussaïd et al. (2012) organized 119 people in France and adopted the same ring corridor as Jelić et al. (2012) with an advanced thinking of bidirectional flow design and allowance of overtaking maneuver [93, 94]. They found the transitions between organized traffic state and disorganized stated and the unstable dynamics arose from the speed variations. Later in China, Kuang et al. (2015) performed a simplified single-file experiments with up to 26 students walking in a circular path with open boundary [95]. From video observation, they discovered the velocity variation, step fluctuation, and the delay from ‘stop-and-go’ waves.

To investigate the flow states under hypercongested situation, Jin et al. (2017) organized up to 278 college students to walk in a ring corridor with an inner width of 2 m and outer width of 4 m [60]. As a combined consideration of Jelić et al. (2012) and Moussaïd et al. (2012), both unidirectional and bidirectional flows were considered and the state transitions were described before and after lane formation process [93, 94]. Also, the fundamental diagrams for both flows were calculated and compared with previous works. Recently, Rahman et al. (2017) established similar experiments with Dias et al. (2014) considering the impact of different turning angles (60°, 90°, and 135°) on walking velocity [87, 96]. Results showed that instead of a linear decrement of speed with angle as discovered in Dias et al. (2014), 90° corridor exhibited the lowest mean speed of the participants. They also found that pedestrians preferred walking at the inner side of the corridor due to the shorter path.

Animal-based turning experiment and its relevancy to pedestrian crowd under panic conditions started from Shiwakoti et al. (2011) [7, 80]. The experiments described the egress of panicking ants from a squared and a circular rooms with an attention on the turning movement effects on egress dynamics. Results showed that turning movements could have negative effects on ants’ outflows. The findings were also compared with a real-life in-store stampede where turning movement occurred at the door and the results showed a consistency between humans and ants in terms of the negative effects of turning movements. They also found that the placement of an obstacle at the exits could increase the flow. Later, Dias et al. (2012) set up a similar experiment with panicking ants egressing from a squared chamber [81]. They also found the flow reduction arose from turning movement that the flow was 20% lower than straight movement. Dias et al. (2013) further set up a more complex environments with turning, merging, and weaving configurations [97]. Escape rates under four angles of turning corridors (30°, 45°, 60°, and 90°) were compared with straight corridor and the flow reduction percentages were -8%, -20%,-16%, and -27%, respectively.

3.3. Egress and Ingress

Egress refers to the movement process of or exiting/outflow of a space while ingress means the opposite action, i.e., entering/inflow of a space. This two processes are also named as “entering and exiting”, “inflow and outflow”, or “inlet and outlet” by some other studies. As shown in Figure 2, egress and ingress process can be interpreted as the combination of straight line and turning movements. Since there is no direction change during egress and ingress process, the two types of movements also belong to the unidirectional flows. Egress and ingress movements often occur at fixed bottleneck sites such as exit and entrance. Space competition resulting from bottleneck geometric restriction is the main cause for the flow reduction.

Given the imperative role of egress process for both daily activities and emergency evacuations, during the past two decades, crowd egress has been one of the most popular topics among all the complex movements. Literatures are quite rich in egress experiments using both humans and animals under normal and emergency situations. Based on the past studies in this topic, the geometric of bottlenecks can be divided into two general categories, i.e., channel or corridor bottleneck and door bottleneck.

In addition to the measurements of the macroscopic and microscopic flow characteristics, the top two research topics for egress and ingress movements are the influence of bottleneck width on capacity, and the performance of obstacles near the exits. Specific flow rate of exit segment is often considered as the benchmark for comparison analysis among different measuring results.

Regarding channel/corridor bottleneck, Daamen and Hoogendoorn (2003)’s experiments series included the egress of crowd into a channel bottleneck with a width of 1 m and 2 m, respectively [56]. They further presented the detailed analysis results in Hoogendoorn and Daamen (2005) [20]. The microscopic flow characteristics for the two bottlenecks were analyzed and compared to the straight corridor without bottleneck. Based on the analysis, the researchers found the self-organized phenomena including the “lane formation” and the “zipper” effect.

With respect to door bottleneck, Helbing et al. (2005) conducted egress experiments under both regular and panic-like escape situations [9]. They examined the effect of placing an obstacle near the exit on the outflow and found the obstacle placement could avoid the clogging effect and increase the flow under both competitive intensities. Likewise, in Japan, Nagai et al. (2006) examined the effect of different movement maneuvers (walking and crawling) on the egress flow through a door [98]. They found the increase in the mean escape time with the number of walker as well as crawlers. It was observed that the mean flow increased with density and saturated at the capacity of the door. Later in China, Zhang et al. (2008) organized 60 students to egress a classroom with fixed exit width of 1.1m [99]. Based on video observation, they calculated the distribution of premovement times, velocity variations, continuous outflow at exit, dislocable queue, and monopolizing exit. They discovered that evacuation times displayed a normal distribution and arrival time and escape order showed linear dependence with similar slope. It was also noticed that coordination among escapers is beneficial to the outflow.

To examine the quantitative relations between the width of bottleneck and the specific flow, Kretz et al. (2006) organized 94 participants (40, 80 persons cases) to walk through a door of varying widths ranging from 40 cm to 120 cm with a stepwise increment of 20 cm [100]. Time headway and specific flow were analyzed for different scenarios. Results showed that the specific flow declined with the increase of width when the door only allowed one person to pass and remained a constant value at larger width. They also compared the flow results with other studies. Similarly, Seyfried et al. (2009) performed controlled experiments controlling the channel width from 0.8m to 1.2m with a stepwise increment of 0.1m [21]. They measured the specific flow under different density levels (20, 40, and 60 pedestrians). Individual velocities, local densities, and time gaps for different width of the corridors were analyzed from the trajectories. They discovered that bottleneck capacity increased almost linearly with width and congestion might occur below maximum capacity.

Likewise, to develop an automatic trajectory extraction software for pedestrians, Boltes et al. (2010) organized 250 pedestrians to walk in two types of bottleneck corridors [67]. Microscopic characteristics such as the spatial and temporal variations at bottlenecks were analyzed later by Liddle et al. (2010) and Liddle et al. (2011) [101, 102]. They stated the reasons behind the flow fluctuations at bottlenecks as a physical effect, i.e., a new stepping function manner that could be only performed at bottleneck.

Later in China, Song et al. (2011) and Tian et al. (2012) carried out similar experiments with Seyfried et al. (2009) considering exit corridor width from 0.5m to 1.4m with a stepwise length of 0.1m [103, 104]. Flow features such as average speeds, group speeds, density-speed-flow relations, and time headway distributions were analyzed. They also compared their specific flow results with Kretz et al. (2006) and found an agreement before 0.7m and large disagreement after 0.7m. In contrast with the previous focus on narrow bottlenecks, Liao et al. (2014) adopted a wide door with up to 5 m width to study the egress flow characteristics [105]. Density-velocity relationships inside the bottleneck were found to be independent with door width. In addition, a linear dependency was discovered between the flow and the bottleneck width.

The optimal design of exit areas also attracted empirical researchers. Sun et al. (2017) conducted controlled experiments in a subway station environment to examine the funnel shape bottleneck design on pedestrian flow operation [106]. Testing different angles of funnel shape, the optimal funnel angle was found between 46° and 65° under all flow conditions. This result was in agreement with the previous empirical evidences on turning angle effects, as the funnel shaped exit design could reduce turning angle and conflicts at exit regions [10].

Thereafter, more influence variables were considered to study the egress capacity at bottleneck especially with regard to the heterogeneity flow. Daamen and Hoogendoorn (2010 and 2012) conducted a series of experiments to examine the door capacities considering different door width (from 50 cm to 275 cm), population composition, stress level, and the presence of door [107, 108]. After a series of statistical tests, except for stress level, all the other parameters passed the significance tests. Later, Garcimartín et al. (2014) and Garcimartín et al. (2016) conducted a series of egress experiments considering door width (0.69m, 0.75m) and competitiveness (low, moderate, and high) [109, 110]. They discovered that the time headways displayed a heavy-tailed distributions while the burst sizes decayed exponential distribution.

Bode et al. (2015) involved 12 pedestrians to egress from a room with 6 exits considering the social group effects on egress time [111]. Their results suggested that the presence of social groups increased egress times. Further, no clear time variation was found between individual and group movements. Similarly, von Krüchten et al. (2016) and von Krüchten and Schadschneider (2017) arranged 32-46 participants (dividing into social groups with 4, 6, and 8 people) to egress a narrow door with widths of 0.8m and 1.2m [112, 113]. Macroscopic and microscopic features of the egress flows were analyzed. They discovered that coordinated social grouping, e.g., queuing at exit and moving in a compact manner, could beneficial to the outflow by reducing the conflicts. This finding also indicated that cooperative compact movements had different effects on the flow operation than the bursts release after clogging. Similarly, Nicolas et al. (2016 and 2017) examined the effects of individual’s selfish and selfishness behaviors on the global flow characteristics by organizing 80 people to egress through a narrow doorway with a 72 cm width [114, 115]. They found that flow increased monotonically with density and the selfish fraction. It was an interesting finding that “selfishness does not mean coordination” because evading behavior could also lag the flow. A tradeoff between the number of evading and overtaking behaviors at the bottleneck that could maximize the flow should be further studied.

Concerning the ingress movement from the outside area into a confined space, only a few of empirical studies could be found. Ezaki et al. (2016) conducted controlled experiments employing a single-file stream of pedestrians successively ingress into a confined room with closed boundary [116]. Collective features, e.g., location and density distribution and individual characteristics such as the choice of location, were analyzed from video recordings. The revealed pattern such as people preferred to select location near corner indicated the significance of psychological and anticipative factors around an individual. The same year in China, Liu et al. (2016) drafted 40 students to reproduce the ingress and egress process in a rectangular room with a 1 m entrance and 1.5m exit [117]. It was discovered that the ingress order had a significant influence on pedestrian’s location distribution in the steady state and the egress order. They also found that inactive pedestrians had a negative effect on the movement as they impeded the flow. Another experiments in Sieben et al. (2017) aimed at reproducing the entrance process of music events [91]. More than 270 people rounding a semicircle room were asked to enter into two narrow exits. It was found that there was a sharp increase of density from 3.8 to 8 Ped/m2 within 10s.

Regarding field data, Cepolina and Tyler (2005) observed the ingress and egress process of metro passengers flowing into the entrance of escalator in London underground station. They discovered the temporal transition of inflow and outflow from the video. In addition, they established the inflow and outflow relationships at a bottleneck with a width of 80 cm. Apart from the individual walking speed in straight movement, Berrou et al. (2007) also observed the ingress process of passengers entering a train station from a high view [74]. Simulation parameters required by crowd simulation software Legion [118], such as the estimate arrival rate and preferred speed, exit flow, and densities, were extracted from the videos and adopted to set up simulation scenarios. Results showed that the simulated flows and densities were in line with empirical data.

Apart from the empirical investigation of human subjects, animals experiment approaches in studying egress movements are also found in the literature.

Most of the animal experiments were carried out using ants. The experiment setups for ants egress experiments were developed to replicate the similar set up for human experiments, i.e., circular chamber [8, 119, 120], squared chamber [121123], and squared chamber with oblique corner near exits [124126]. The common parameter measured is the escape flow at exits. Animal models have been the popular choice to study stressful condition like panic behavior as ethical considerations prevent creating a panic-like situation in pedestrian crowd situation [7].

Altshuler et al. (2005) established a circular chamber with two symmetrically located exits to examine the symmetry breaking phenomena in panicking ants [119]. It was discovered that under panic state, the use of two exits were asymmetrical while the use of exits displayed symmetric under normal situation. Later, Shiwakoti et al. (2009), Shiwakoti et al. (2010), and Shiwakoti et al. (2011) conducted ants experiments to study the effect of exit location and obstacle placement on outflow of panicking ants [7, 120, 127]. It was found that corner exits tend to have higher flow compared to middle exits in squared chambers. Also, the placement of obstacle could in general increase the flow as compared with column free situation. Burd et al. (2010) followed Shiwakoti et al. (2009)’s experiments to study ants’ nested behaviors in normal state [8]. They also found the obstacle could increase the egress flow of normal ants. Soria et al. (2012) found the ‘faster-is-slower’ effect in ants traffic under various stress levels [124]. Their results also exhibited no selfish traffic behavior among ants. Likewise, Boari et al. (2013) observed the consistence finding in terms of the lack of selfish behavior [125]. However, ‘faster-is-slower’ effect was not found in their result. Later, Parisi et al. (2015) further analyzed Soria et al. (2012)’s data and carried out a discussion with human traffic rules [128].

Recently in China, Wang et al. (2015) used a single-exit chamber with different exit widths to perform ants egress experiments under stressed situations [121]. They found that the average flow was independent linearly on the exit width which was in contrast with human behaviors. Wang et al. (2016) further built a two-exit chamber to study ants egress through exits with the change of exit location and spacing [122]. It was found that the most efficient exit layout was the longest spacing of the two exits and this finding was correspondent with Shiwakoti et al. (2013) [123]. Symmetric breaking was also observed among stressed ants along with Altshuler et al. (2005). They also discovered the different performance of Social Force Model with respect to describing humans and ants movement in positions analysis, density map, velocity direction, and trajectories.

Another insect that has been used to study crowd collective movement is woodlice. Sobhani et al. (2014) and Sobhani et al. (2015) placed 120 panicking woodlice in a squared chamber with a 1 cm-width middle exit [129, 130]. Densities in near exit region and the flow of the exits were measured based on video analysis and then the relationship between jam density and exit capacity and the fundamental diagrams for near exit region were discovered. They found that the fundamental diagrams for animals in panic status had some variations with the normal situation. And as the rise of stress level among woodlice, the number of blockage near the exit increased.

Apart from ants and woodlice experiments, several researchers used mice to study the egress movement under stressed conditions. Saloma et al. (2003) carried out an experiment with 60 mice egressing from a water pool onto a dry land through door bottleneck with different widths and separations [131]. It was observed that mice behavior was to some extent similar to the simulated results from pedestrian models. Self-organized queuing could be observed at the width that could only allow one mice to pass at the same time. As the door width increased, the self-organized pattern was broken due to the space competition. Saloma et al. (2015) further analyzed their experiments data from the aspects of individual training and crowded degree [132]. They found that prior training of individual mouse could increase the escape efficiency compared with the untrained situation. The training effects were more significant under more crowded situation.

Later in China, Lin et al. (2016) purchased a group of 95 female mice with a uniform body size to perform experiments in a dry rectangular ground with 2 cm width middle exit [133]. Different number of Joss-sticks was burned into smoke to stimulate the mice escape at various level of panic or stress. It was observed that egress time and the number of clogs increased with the levels of stress. It was concluded that the selfish competitive behavior displayed by mice was the main cause for the increase of egress time. Lin et al. (2017) further improved their experiment setting with a consideration of obstacle placement distance and the partition of near exit regions [134]. It was observed that the placement of obstacle could increase the flow by a maximum of 36% and 26% for regular exit region and partitioned exit region, respectively. The same group of researchers further analyzed the effects of exit width and position on mice egress movements. Chen et al. (2017) found that enough exit width enabling two mice pass the exit side by side could avoid ‘faster-is-slower’ effect [135]. Further, Chen et al. (2018) found that ‘faster-is-slower’ effect was less likely to occur at corner exits than middle exits [136]. In another mice egress experiments, Oh and Park (2017) examined the influence of the exit angle on egress dynamics with a group of 50 mice [137]. They found that, in general, mean velocity and total egress time decreased with angle.

Regarding larger animals, Zuriguel et al. (2014) studied the egress dynamics by observing the feeding process of sheep flock passing through a narrow fence with in a farm [138]. They found that the time headway of sheep flock exhibited a power-law distribution and the burst sizes obeyed an exponential decay. To further investigate the door width and obstacle effects, Garcimartín et al. (2015) compared the effect of 77 cm and 94 cm doors in with and without obstacle situations [139]. By examining the mean egress time per animal for the four scenarios, it could be found that the most efficient setup was the 94 cm door with obstacle and the least case was the 77 cm door without obstacle. Zuriguel et al. (2016) further studied the effect of obstacle position on the egress flow [140]. They found obstacle position had an influence on the sheep flow and 80 cm was the best distance for the 77 cm door.

3.4. Opposing

Multidirectional movements are more complex compared with unidirectional movements due to the increase possibilities of interactions. The simplest case of multidirectional flows is usually called ‘bidirectional’ flow where movements are only operated in two directions.

Opposing movement refers to the situation when two streams of people coming from two opposed directions confront at a confined region. It can be also named as counterflow adopted by many studies [141145]. Because there can only be two direction of flows in opposing movements, in this case, opposing can be regarded as a special scenario of bidirectional flows (or head-on weaving) where the two streamlines interact with an angle of 180°.

Opposing movements can be frequently observed in the daily life such as the boarding and alighting process in a crowded metro station, the movements on bridge during special events [146, 147]. In addition, opposing movements have been identified as the cause of several crowd disasters around the global occurring from the early 1980s to the late 2010s [3, 32].

Given its imperative role in practice, due amount of research efforts has been put into the study of opposing movement which include many empirical evidences.

Several Japanese researchers conducted opposing experiments with an initial purpose to validate the lattice gas model proposed by Tajima and Nagatani (2001) [148]. Isobe et al. (2004) organized up to 70 people to walk from two opposed sides of a rectangular corridor [141]. They presented the evolutions of arrival time and mean velocity as the increase of density. They found that, as the rise of density, the arrival time increased but the mean velocity decreased. Nagai et al. (2005) performed similar experiments with both walkers and crawlers in the opposing case [142]. The arrival time of crawler was dependent on the initial location and the mean arrival time increased with density. In addition to the straight line scenario, Asano et al. (2007) also included an opposing scenario. They illustrated the headway-speed relationships of opposing flows and found that mean speed tend to decrease in shorter headways.

In Europe, controlled experiments of opposing movements were first conducted by Daamen and Hoogendoorn (2003) [56]. Their experiment series also included the case of opposing movements. They applied the similar configuration of their previous straight line and egress scenarios including the desired speed and density levels. However, they only provided the qualitative analysis results for the bottleneck scenarios. Likewise, Helbing et al. (2005)’s experiments sets also contained the opposing movements in a corridor with a short and a long bottlenecks [9]. They presented the flow values and time gaps distributions for the two opposing cases and results showed that long bottleneck was worse than short bottleneck in terms of the effects on opposing movements. Later, Kretz et al. (2006) carried out an opposing experiment with 67 participants in a 2 m width corridor [143]. They calculated the passing times, speeds, and flows and found that the sum of flow in opposing flow was larger than the straight line flow. They also investigated the lane formation and symmetric breaking in opposing flows and presented the frequency of number of lanes in the corridor.

Moussaïd et al. (2009) performed laboratory experiments (40 participants) and field observations (2670 samples) including opposing movement [149]. In their findings, individual’s side preference in the evading maneuvers occurring in opposing movement was interpreted as a behavioral coordination behavior as a result of cultural differences. In Dutch, Versluis (2010) conducted a series of laboratory experiments considering age, body size, gender, free speed, travel purpose, maneuverability, pedestrian number, and predicted postencroachment time [150]. They investigated the lateral and longitudinal evasion of pedestrians with a mean value of 0.26m and 0.11m/s for opposing case. Back to Germany, Seyfried’s team also collected datasets for opposing movement. Zhang et al. (2012) separated the opposing flows into stable separated lanes and dynamical multilanes flows with balanced and unbalanced flow ratio based on different ordering levels. They painted the fundamental diagrams for each flow types with different corridor width and compared the results with their previous straight line movement data [151]. Results indicated that ordering levels and density measurement methods did not significantly affect the fundamental diagrams. They also compared the maximum flow values of bidirectional data (1.5 Ped/m/s) with their previous unidirectional data (2.0 Ped/m/s). In addition to unidirectional flow, Cao et al. (2017) also reported the experiment results of opposing movements in a corridor with a width of 4 m [13]. Under the density level of 1.5Ped/m2, the estimated specific flow value for opposing movements was 1.1Ped/m/s. This value was observed to be significantly lower than the unidirectional movement.

To investigate the impact of social group and flow ratio on opposing movement, Gorrini et al. (2016) organized controlled walking experiments in Japan [42]. Their results suggested that the rise of flow ratio had negative impact on the speed of pedestrians. Walking in groups could slow the speed because of the difficulty in moving coordination among group members. In contrast, coordination effects were also found to be positive in opposing movement. To explore the experimental evidence in the “gridlock” effect existing in CA models when simulating the opposing movement, Xue et al. (2017) conducted comparative experiments in both discrete and continuous space [152]. Through the time step analysis of pedestrians’ position as well as the global density and specific flow, gridlock was not observed even under high density opposing flow. This result demonstrated the remarkable collaboration behaviors of pedestrian group movement especially at congested situation.

Recently, opposing movements were also investigated in animal systems. Wang et al. (2018) observed the bidirectional movement characteristics for nested ants with loaded and unloaded status through a less competitive experiment [153]. They measured the speed for unloaded and loaded ants, distance headway distribution over speed, and the spatial-temporal analysis of encounter behaviors through trajectories. Results showed that ants appeared to be equally sensitive to distance headway in both uni- and bidirectional flows. Head-on encounter behaviors with the following ants could reduce movement efficiency. Unloaded ants tend to spend more time in communicating with encountering ants than the loaded ones. However, quantitative comparison with vehicular and pedestrian traffic was not conducted.

3.5. Weaving and Intersecting

In opposing flows, the interacting angle of the two streamlines remains a constant value of 180°. Past studies have highlighted that oblique interacting angle of different streamlines can also have a significant impact on the crowd movements [12, 87, 150].

In this study, weaving movement refers to the situation when two streams of pedestrian flows arrive at a fixed region (such as an intersection) at the same time and they can only keep their walking direction by cross over the other steam (please refer Figure 1). It is a commonly observed type of multidirectional movement as more than two numbers of streams may involve in the interactions. Weaving has several alternative forms such as intersecting and crossing, and if the number of involving streams is only 2, it can be also categorized into bidirectional movement.

The first empirical study on weaving movement for pedestrians was found in Daamen and Hoogendoorn (2003) [56]. However, no empirical analysis results for opposing flows were presented in the study. Later, Helbing et al. (2005)’s experiment sets also contained a right-angle weaving scenario with two stripes of flows [9]. Compared to their straight line data, they found the stripe formation phenomenon as there was a significant irregularity and clusters in the passing time. Asano et al. (2007) further analyzed the speed-density relationships (Speed=A+BDensity) under the impact of interacting angle [12]. They found that parameter A was significantly influenced by several angles (0°, 45°, and 90°) while parameter B was not influenced by the angles except for the right-angle case. Apart from the opposing movement reviewed in the above section, Versluis (2010) also considered oblique interacting angle including 45°, 90°, and 135°. It was found that the angle exhibited a positive effect on mean interaction point and showed a negative effect on mean longitudinal evasion. Similarly, Plaue and Chen (2011) organized two groups of people (54:46) to participant a right-angle weaving experiment [154]. They applied a crowd density estimation method to analyze the trajectories extracted from video. It was found that the nearest-neighbor kernel density method could provide a reasonable estimate of the density of human crowds. Back to Seyfried’s new experiments, Cao et al. (2017) also reported the results of weaving scenarios including the right-angle bidirectional and four-directional weaving flows [13]. Similarly, at the density level of 1.5Ped/m2, the estimated specific flow for bidirectional and four-directional weaving flows was 1.2m/s and 1.1m/s.

In Hong Kong, Wong et al. (2010) performed controlled experiments considering five intersecting angles (0°, 45°, 90°, 135°, and 180°) [155]. The aggregate empirical data were used to calibrate a model and the results showed highly significance. Adopting the calibrated model, they compared the mean speed transitions with reference stream under different density ratio and total densities. The researchers found that as the rise of angle, the speed for reference stream decreased. Later, Wu and Lu (2013) conducted weaving experiments considering the dimension of passageway and flow ratio [156]. They defined a pedestrian weaving zone (PWZ) and discovered the flow characteristics within PWZ. They found that the flow values and flow ratios were more concerned with PWZ instead of the entire walking region. They also established a utility function to assess the PWZ operations based on three novel indices, i.e., weaving intensity, trajectory offset ratio, and density distribution factor of weaving points. Likewise, Sun et al. (2014) performed controlled experiments with 50 students to understand the mechanism of weaving movement at congested situations. Along with the flow characteristics of weaving movements, they also found that weaving tend to occur within an area which was in line with Wu and Lu (2013). Later, Lian et al. (2015) performed a four-directional intersecting experiments with 364 young Chinese students [157]. They examined the effects of side preference on lane formation and the placement of obstacle on the flow stability. It was discovered that lane formation process could be faster when students selected right-hand lane in the corridor. In addition, the placement of obstacle in the middle of the intersection could stabilize the intersecting flows. Sun et al. (2017) and Sun et al. (2018) further investigated the impact of intersecting angles and obstacle effects on intersecting movements [149, 158]. It was found that the effects of angles on flow speed vary at different flow conditions. But the effects of the size of the circular obstacle were monotonous with the speed increment and acceleration dispersion.

Animal-based interesting studies could be only found in the ants experiments conducted by Dias et al. (2013) [97]. In addition to the turning configuration, they also set up a junction corridor with right angle that enabled the bidirectional weaving movements. They discovered the transitions between the two weaving streams; i.e., one moving stream could block the other stream.

3.6. Merging and Diverging

Merging movement can be regarded as the combination of turning and weaving movement. It can be frequently observed at angled corridors/passageways such as T-junction or the stair-floor interface of a high-rise building. Moreover, merging configurations have also been identified as the causes for several stampede accidents including the famous Love Parade disaster [3, 32]. Due to the multiplication of the interactions among turning and weaving movements, merging movements are more complex.

Empirical evidences on merging movements were not sufficient before the year of 2014 [4, 58, 159, 160]. There was a burst of empirical studies on merging behaviors in the most recent four years [3, 161168].

Boyce et al. (2012) examined the merging behaviors at stair-floor interfaces by performing evacuation drills with 581 participants in three buildings. They noted that the location of stair-floor connection points and the population composition could have a potential influence on merging flow patterns. The same year, Ma et al. (2012) conducted a similar evacuation drills in ultra-high-rise building [169]. They also observed that the merging behaviors at stair-floor interface could impede the flow of evacuees. To replicate the emptying process of mass gathering events in a stadium, Burghardt et al. (2013) performed controlled experiments in the exit of a football stadium where pedestrians moved from spectator seats upon receiving the signal and successively formed three stable streamlines merging into the outlet doors [86]. They only presented the fundamental diagrams of the downstream stairs; details on the merging characteristics were not analyzed. More recently, Huo et al. (2016) performed evacuation drills with 73 participants in a 9-floor high-rise building [161]. They discovered the negative effect of merging behavior on total evacuation time and speed through quantitatively comparison.

With respect to the merging flows on ground surface, Zhang et al. (2011 and 2012) performed merging experiments at T-junctions and compared the fundamental diagrams of merging with straight walking [58, 84]. After deriving the fundamental diagrams of merging flows, they found an increase in speed within after merging areas compared to before merging regions. Shiwakoti et al. (2015) studied the impact of merging angle and desired speed on the microscopic characteristics of merging flows in before and after merging areas. They found that, as the rise of merging angle, the reduction in speed increased. Similarly, Aghabayk et al. (2014 and 2015) also investigated the crowd merging behaviors considering merging angles and desired speed with asymmetric design of corridor [162, 163]. They found that the arrival and departure flow increased with speed and time headways decreased with speed. It was observed that merging angle was also influential to flow and time headways which was in consistent with the findings from Shiwakoti et al. (2015).

Later, Shahhoseini et al. (2016) conducted ants experiments to further understand the merging angle effects (60°, 90°, and 120°) [164]. They found the negative effects of merging angle in ant trials in panic situation. Further, Shahhoseini et al. (2017) combined the factors of merging angle and symmetrical configuration and designed 5 types of merging corridors to perform laboratory experiments with human subjects [170]. It was discovered that the merging behaviors of ants and humans displayed similarly effects arising from differences in architectural configurations.

Same year in China, Lian et al. (2017) established a merging configuration of branched flow merged into main flow with a further consideration of changing the width of branched corridor (0.8m, 1.6m, and 2.4m) [165]. It was observed that the merging ratios had a significant influence on the operations of upstream and downstream of merging areas for pedestrians flow. In the same year, Cuesta et al. (2017) performed merging experiments in a mock-up tunnel with 77 participants to reproduce the merging process of passengers exiting from the train and pedestrians walking on the platform [171]. They used the distribution of instantaneous specific flow to represent the flow dynamics. It was found that merging section had negative effects on the outflow of people at rail tunnel in terms of specific flow.

As illustrated in Figure 2, diverging movement can be defined as a reverse process of merging movement, where a united stream of pedestrians separated into multiple streams with different destinations. Different from the fluctuating flows near a bottleneck, the flow states are stable in both before and after diverging.

A few empirical studies on exit choice behaviors could be categorized into diverging movements. Haghani et al. (2016) and Haghani and Sarvi (2017) performed exit choice experiments considering various exit number (2, 3, and 4) and exit position and exit width (50cm, 100 cm) [172, 173]. Instead of movement characteristics, their focuses were in the decision-making aspects. In another exit choice experiments, Wagoum et al. (2017) reported the flow features, e.g., flow, density relationships of the participants at exits. It was found that under normal situation individual’s choice of exit only was dependent on the shortest distance while under higher density situation the load balancing for two exits could be observed.

3.7. Random Flow

Researchers have tried to reproduce the random flow situations by conducting experiments with multiple crossing flows. Dyer et al. (2008) set up a circle with sixteen equally angle-divided directions on the edge [174]. People were asked to stand inside at the center of the circle and individuals would be informed with a direction to walk outside the circle. They tested the role of prior information of walking direction and the conflicting directional information on the group directional decision-making. However, the flow characteristics of the experiments were not analyzed as the researchers were more interested on the psychological aspects of the participants. However, it is to be noted that there is a greater need of empirical data to predict the walking directions and potential conflicts due to random flow.

Researchers from computer vision field have also contributed to the empirical understanding of random flow. Zhou et al. (2012) analyzed the collective behaviors a crowd of pedestrians in a train station walking in random flow directions [175]. Using a mixture model of dynamic pedestrian agents, the future trajectories of pedestrians were predicted based on the observed past trajectories. Also, they classified the random walkers into various collective motion patterns and visualized them in different clusters.

3.8. Summary

Table 1 shows the summary of key findings and recommendations from the literature that conducted empirical studies on the externally governed complex multidirectional pedestrian traffic movements. The table shows the source of the literature, the movement type, the type of infrastructure considered (walking facility, bottleneck, and obstacle), subjects considered (pedestrians, animals), controllability of the variables (controlled laboratory experiments, field observation), and competitiveness among pedestrians (normal, emergency/evacuation drills, and panic). Further, the table provides a summary of the key findings, limitations, and recommendations for future studies and further use of the data generated from a particular literature in other studies. The filled dot in the table shows what approach has been followed in a particular literature while empty cell shows missing approach in a particular literature.

Table 1: Summary of empirical results on externally governed complex pedestrian traffic movements.

From the table, it is clear that initial studies have focused mainly on the empirical studies of unidirectional pedestrian flow, mostly at the corridor or at the bottleneck like door. Recent empirical studies have focused more on the complex multidirectional movements like weaving, crossing, and merging/diverging. Further, most of the studies are conducted under normal walking conditions. There have been increasing interests on the pedestrians’ multidirectional movements under emergency or panicked conditions in recent time. As such, there has been rise in the study of animal models, particularly their relevancy to pedestrian traffic under emergency conditions.

In the next section, we present a discussion on the implications of these findings and directions for future research.

4. Discussions

Empirical data of complex pedestrian crowd movements are playing increasing imperative role in the verification and validation of crowd simulation models, design and planning of pedestrian walking facilities, scheme and evaluation of evacuation plans, development of image processing and visualization tools, coding of game engines, and trajectory planning for robots as well as the understanding of the nature of collective traffic phenomena in biology systems. Therefore, as observed from the literature review, there has been surge on empirical data collection and analysis over the past two decades. From the synthesis of the findings from these literatures, we have identified some opportunities and challenges as described in the following sections.

4.1. Anticipation of Future Empirical Data Requirements

To evaluate the current work and identify the research gaps based on the reviewed empirical studies, we utilize the data in Table 1 based on the number of studies and their belonging to each category. The summary of the statistics is presented in Figure 3.

Figure 3: Summary statistics from Table 1. (a) Histogram diagram illustrating the distributions of the quantity and proportion of empirical studies on different multidirectional movements over different time periods. (b) Tree map to compare the proportion of the empirical study that focused on competitiveness, controllability, subjects, and geometric conditions.

Figure 3(a) summarizes the number of studies conducted over three time periods (2003-2007, 2008-2012, and 2013-2018) under each movement type. From the general trend of column height showing in Figure 3(a), we can notice that egress and straight line movements are the most popular topics among empirical studies while diverging and random flows are less studied through empirical approach. Also, from the distribution of studies, we can observe that earlier studies (2003-2007) in unidirectional flow have large proportion of straight line and egress experiments while multidirectional flow contains study mostly on the opposing and weaving movements. Meanwhile in the recent decade (2008-2018), researchers not only continue to explore the empirical evidences in straight line and egress movements, but also started to collect empirical data on turning, merging, and ingress movements. The emergence of increased studies on these movement patterns may arise as a result of several recent crowd disasters that involved turning and merging movements [32]. It can be further noticed that the dynamic features of diverging and random flows still require further research attentions.

In addition to the flow-based summary, we also present the comparison of contextual-based statistics of the reviewed studies in a set of nested rectangles as shown in Figure 3(b). The hierarchy of data generated from Table 1 in terms of the geometric (bottleneck or obstacle), investigating subject (human or animal), controllability (laboratory experiment or field observation), and competitiveness (normal or emergency or panic status) is presented in Figure 3(b).

Overall, laboratory experiments occupy the predominant proportion compared with field observations. In addition, the majority of empirical evidences are in human subjects while it is minor in biological entities. Regarding the study on competitive behavior, studies are mostly conducted under normal status but less in emergency and panic situations. Therefore, in future, empirical studies that can improve our understanding on mechanisms of multidirectional movements under emergency situations need to be investigated.

4.2. SWOT Analysis for Empirical Data Collection Approaches

During the review process, we find that the data collection methods can have substantial influences on the obtained results. To help researchers select the suitable approach to satisfy their data requirements, assessments of different data collection approaches in terms of their underlying internal advantages and disadvantages as well as the potential external factors that can exploit or obstruct the advantages are needed.

Therefore, we perform comparative SWOT (Strengths, Weaknesses, Opportunities, and Threats) analysis for different data collection methods to identify the positive and negative prospects and classifying them into internal (Strengths and Weaknesses) and external (Opportunities and Threats) factors [195].

We conducted a first set of SWOT analysis to examine the variations of controllability of experiments between laboratory experiments versus field observations.

The utmost advantage for laboratory experiments is the strong controllability towards investigating targets that satisfied the straightforward requirements. Researchers can set up the predesigned walking environment, select the satisfying participants, choose the intended data measurement techniques, or even define the manners of movements by giving instructions or pressing forces to the participants. Through such purpose-oriented approach, the specific influencing factors of the crowd movements can be examined one by one in a flexible manner and in detail. Therefore, empirical data from laboratory experiments should be the suitable materials for the development of microscopic models. However, the benefits of the well controllability for laboratory experiments may have to tradeoff with the risk of not completely representing the real-life situation. For instance, differences between the pedestrian movements during emergency escape and laboratory experiments have been quantitatively compared in Shiwakoti (2016) [196]. It is suggested that the physiological aspects of pedestrian should be further examined such as the effect of peoples' consciousness of their involvement in an experiment on their performed behaviors [197]. Moreover, not all the factors are controllable especially in animal experiments. This is perhaps the right reason for the absence of opposing experiments using biological entities. Further, due to the simplifications of the experiment setups, the effects of latent uncontrolled variables can also affect the global results and such effects might be amplified due to the control of certain variables. Therefore, in the future, in-depth statistical analysis should be conducted to examine the interactions of manifest and latent variables of pedestrian dynamics by using more comprehensive empirical datasets. Nevertheless, the advancements made on data collection and analysis techniques as well as optimal experimental design methods may help increase the size of the database, improve the data accuracy of experiments, and reduce the time and labour costs of the researchers.

With regard to field observation, the major pros for this approach are the authenticity of data quality and the relative lower baseline requirement to carry out the field observation. Field data can represent the real-life circumstance giving no other effects such as the subjects’ awareness of being monitored. In addition, it can be easily implemented, e.g., placing a camera at a fixed location or even using the exiting videos shot by surveillance devices. However, there are limitations of field observation as well. Frist of all, not all the areas or contexts are permitted to conduct field observation research because of privacy and security reasons. This point is particularly suitable to the unmanned aircraft vehicle (UAV) as many countries have banned UAVs and drones in their restricted airspaces. Second, it usually takes a long time period or requires switching various sites to capture the intended contexts and obtain the adequate samples. Fortunately, the emerging of smart cities and big data analytics provide opportunities for large field data collection from multiple sources. Recent technological advances in the imaging-sensing industry and the machine learning research community have made previously inaccessible data largely available. The imaging industry enables high-resolution (1080p) CCTV cameras to perform crowd monitoring. Therefore, real time field data collection and short term analysis of pedestrian crowd have been implemented in many cities [198]. Meanwhile, the ubiquity of smartphones has also offered various opportunities for collecting pedestrian traffic data in the real-life fields [199]. Consequently, the relating issues on personal privacy and public information security should also be the threats for this approach.

The highlights of the SWOT analysis for laboratory experiments and field observations are summarized in Table 2.

Table 2: SWOT analysis for laboratory experiments versus field observations.

We further conducted a second set of SWOT analysis between animal-based approach versus human-based approach.

The foremost strength for animal-based approach is identified as the possibility of conducting real panic experiments using biological entities. Further, animal behaviors can be more easily captured through laboratory experiments or field observations with lower cost than human approaches. Although there exists obvious discrepancy between animal movement and human motion, the empirical data in panicking animals are also more ‘life-like’ than many mathematical models [36, 196]. There is opportunity to use the animal models to further improve our understanding of collective motion. For example, the communication and coordination mechanisms of ants can be further examined to develop the methods for controlling human crowd [127].

In terms of human-based approach, apart from the most apparent advantage of ground truth data, we also highlight the advantage of providing instructions to pedestrians as the secondary pro in conducting experiments with human subjects. However, human data show large contextual dependency arising from both internal and external factors such as heterogeneity and behavioral uncertainty. Therefore, future empirical databases should be enriched with more psychological aspects [196]. In addition, one major con for human approach is the prohibition of performing dangerous panic experiments due to the ethical and safety reasons. However, the emerge of virtual interactive techniques such as augmented reality (AR) and virtual reality (VR) can help examine human behaviors in more extreme scenarios including the panic escape [200]. Also, researchers should also explore risk free or risk immigration experimental design methods to extend the research scopes to extreme behaviors.

The summaries of the SWOT analysis for animal-based approaches versus human-based approaches are presented in Table 3.

Table 3: SWOT analysis for animal-based approaches versus human-based approaches.
4.3. Future Global Collaborative Efforts on Empirical Data Collection

In the future, in order to accelerate the path of crowd dynamics research, a comprehensive open-source empirical database on root of the different external complex movement patterns and internal influence variables should be established. To realize this goal, interdiscipline collaborations should be encouraged. Some of the researchers from various disciplines have made their datasets available to the public such as open videos and trajectories datasets of controlled laboratory experiments [201], pedestrian crossing behaviors on street [202], field datasets of pedestrian’s walking behaviors in train station, school forum or gallery [203205], and collective motion databases for computer vision analysis [206]. All the above datasets contain pedestrian trajectories and some of the researchers have also uploaded the original videos.

Therefore, mutual assistance and cooperation should be encouraged between research groups across disciplines and across regions around the world. A data-sharing platform should be established so that all researchers working in this field can contribute their datasets.

During our review study, we noticed that many empirical data are not well recorded and have little possibilities for further usage due to the limitations of experiments setups or measuring methods. Therefore, a common standard guidance on performing controlled experiments and field observations should be developed. Further, automatic image processing tools that can restore the trajectories from the published papers can be developed in future to make full use of the existing empirical studies.

Further, researchers working in this field should come together to establish a series of quantitative standardization methods to normalize the empirical data [207]. Also, a comprehensive guideline and standard for performing controlled laboratory experiments are required.

5. Conclusions

Collective motion of pedestrian traffic is sophisticated under complex environments. To better understand the underlying mechanism of collective crowd movements, a lot of empirical studies were carried out through controlled laboratory experiments and field observations with humans subjects and nonhuman biological entities under various scenarios. However, a comprehensive review of the detailed classification of these empirical works from the aspects of complex movements was absent in the literature.

To this end, in this review article, we first categorized the complex movements of pedestrian crowd into two general categories, i.e., external governed movements and internal driven movements based on the governing factors of the motion pattern. Further, considering the hierarchy of movement complexity, we decomposed the external governed movements of pedestrian traffic into several motion based patterns including straight line, turning, egress and ingress, opposing, weaving, merging, diverging, and random flows. Then, we reviewed the related literatures under each motion based pattern with a focus on the complex movement characteristics. Further, a series of evaluations were performed towards each reviewed empirical study in terms of the geometrics setups (whether contained bottleneck or obstacle), investigating subjects (humans or animals), controllability (laboratory experiments or field observations), and competitiveness (under normal, emergency, or panic situations). Moreover, the key findings, limitations, or future recommendations and the further usage of the empirical data in other studies were summarized in a table.

Summary statistics from the aggregated table showed that empirical data were highly rich in straight line, egress flow, and medium rich in turning, merging, weaving, and opposing, but poor in ingress, diverging, and random flows. Researchers were more concerned with human subjects than animal subjects and laboratory experiments were more preferred than field observations. Empirical data were mostly collected under normal situations but less in emergency and panic conditions. Likewise, the studies on bottleneck effects were more popular than the obstacle effects.

We further presented the comparative SWOT analysis for different data collection approaches that included two comparisons: laboratory experiments versus field observation and animal-based approaches versus human-based approaches. At last, we put emphasis on the need for the future global collaborative efforts on data sharing and developing guideline for performing controlled laboratory experiments for the pedestrians complex movements.

Conflicts of Interest

The third coauthor Dr. Shiwakoti has conflicts of interest with researchers from University of Melbourne.

Acknowledgments

This research is sponsored by National Science Foundation of China (no. 71701108) and National Science Foundation of Zhejiang Province (no. LQ17E080007). The third coauthor would like to acknowledge the funding received from ARC Linkage Project LP120200361 for his contribution in this paper.

References

  1. X. Shi, Empirical Investigation on Pedestrian Complex Movement Characteristics and Interaction Behaviors [Ph.D. thesis], Southeast University, 2018.
  2. X. Shi, Z. Ye, N. Shiwakoti, and Z. Li, “A Review of Experimental Studies on Complex Pedestrian Movement Behaviors,” in Proceedings of the 15th COTA International Conference of Transportation Professionals, CICTP 2015, pp. 1081–1096, American Society of Civil Engineers, Reston, VA, USA, 2015. View at Publisher · View at Google Scholar
  3. X. Shi, Z. Ye, N. Shiwakoti, D. Tang, C. Wang, and W. Wang, “Empirical investigation on safety constraints of merging pedestrian crowd through macroscopic and microscopic analysis,” Accident Analysis & Prevention, vol. 95, pp. 405–416, 2016. View at Publisher · View at Google Scholar
  4. E. R. Galea, G. Sharp, and P. J. Lawrence, “Investigating the representation of merging behavior at the floor—stairs interface in computer simulations of multi-floor building evacuations,” Journal of Fire Protection Engineering, vol. 18, no. 8, pp. 291–316, 2008. View at Publisher · View at Google Scholar · View at Scopus
  5. D. Helbing, I. Farkas, and T. Vicsek, “Simulating dynamical features of escape panic,” Nature, vol. 407, no. 6803, pp. 487–490, 2000. View at Publisher · View at Google Scholar · View at Scopus
  6. R. Escobar and A. De La Rosa, “Architectural design for the survival optimization of panicking fleeing victims,” in Advances in Artificial Life, pp. 97–106, 2003. View at Publisher · View at Google Scholar
  7. N. Shiwakoti, M. Sarvi, G. Rose, and M. Burd, “Animal dynamics based approach for modeling pedestrian crowd egress under panic conditions,” Procedia - Social and Behavioral Sciences, vol. 17, pp. 438–461, 2011. View at Publisher · View at Google Scholar
  8. M. Burd, N. Shiwakoti, M. Sarvi, and G. Rose, “Nest architecture and traffic flow: large potential effects from small structural features,” Ecological Entomology, pp. no–no, 2010. View at Publisher · View at Google Scholar
  9. D. Helbing, L. Buzna, A. Johansson, and T. Werner, “Self-organized pedestrian crowd dynamics: experiments, simulations, and design solutions,” Transportation Science, vol. 39, no. 1, pp. 1–24, 2005. View at Publisher · View at Google Scholar · View at Scopus
  10. N. Shiwakoti, M. Sarvi, and M. Burd, “Using non-human biological entities to understand pedestrian crowd behaviour under emergency conditions,” Safety Science, vol. 66, pp. 1–8, 2014. View at Publisher · View at Google Scholar · View at Scopus
  11. N. Shiwakoti, X. Shi, Z. Ye, Y. Liu, and J. Lin, “A comparative study of pedestrian crowd flow at middle and corner exits,” in Proceedings of the Australasian Transport Research Forum 2016, 2016, http://atrf.info/papers/2016/files/ATRF2016_Full_papers_resubmission_147.pdf.
  12. M. Asano, M. Kuwahara, and S. Tanaka, “Multi-directional pedestrian flow model based on empirical data,” in Proceedings of the 11th World Conference on Transport Research, 2007, http://trid.trb.org/view.aspx?id=877133.
  13. S. Cao, A. Seyfried, J. Zhang, S. Holl, and W. Song, “Fundamental diagrams for multidirectional pedestrian flows,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2017, no. 3, p. 033404, 2017. View at Publisher · View at Google Scholar
  14. M. Haghani and M. Sarvi, “Crowd behaviour and motion: Empirical methods,” Transportation Research Part B: Methodological, vol. 107, pp. 253–294, 2018. View at Publisher · View at Google Scholar
  15. J. K. K. Yuen and E. W. M. Lee, “The effect of overtaking behavior on unidirectional pedestrian flow,” Safety Science, vol. 50, no. 8, pp. 1704–1714, 2012. View at Publisher · View at Google Scholar · View at Scopus
  16. Y. Zhao, T. Lu, M. Li, and L. Tian, “The self-slowing behavioral mechanism of pedestrians under normal and emergency conditions,” Physics Letters A, vol. 381, no. 37, pp. 3149–3160, 2017. View at Publisher · View at Google Scholar
  17. A. U. Kemloh Wagoum, A. Tordeux, and W. Liao, “Understanding human queuing behaviour at exits: an empirical study,” Royal Society Open Science, vol. 4, 2017. View at Publisher · View at Google Scholar
  18. M. Moussaïd, N. Perozo, S. Garnier, D. Helbing, and G. Theraulaz, “The walking behaviour of pedestrian social groups and its impact on crowd dynamics,” PLoS ONE, vol. 5, no. 4, Article ID e10047, 2010. View at Publisher · View at Google Scholar · View at Scopus
  19. S. P. Hoogendoorn and W. Daamen, “Self-organization in walker experiments,” in Traffic Granul. Flow ’03, p. 11, 2004. View at Google Scholar
  20. S. P. Hoogendoorn and W. Daamen, “Pedestrian behavior at bottlenecks,” Transportation Science, vol. 39, no. 2, pp. 147–159, 2005. View at Publisher · View at Google Scholar · View at Scopus
  21. A. Seyfried, O. Passon, B. Steffen, M. Boltes, T. Rupprecht, and W. Klingsch, “New insights into pedestrian flow through bottlenecks,” Transportation Science, vol. 43, no. 3, pp. 395–406, 2009. View at Publisher · View at Google Scholar · View at Scopus
  22. D. J. Low, “Following the crowd,” Nature, vol. 407, no. 6803, pp. 465-466, 2000. View at Publisher · View at Google Scholar
  23. D. Helbing, I. J. Farkas, and T. Vicsek, “Freezing by heating in a driven mesoscopic system,” Physical Review Letters, vol. 84, no. 6, pp. 1240–1243, 2000. View at Publisher · View at Google Scholar · View at Scopus
  24. D. Helbing, A. Johansson, and H. Z. Al-Abideen, “Crowd turbulence: the physics of crowd disasters,” 2007, https://arxiv.org/abs/0708.3339.
  25. X. Yang, Z. Wu, and Y. Li, “Difference between real-life escape panic and mimic exercises in simulated situation with implications to the statistical physics models of emergency evacuation: The 2008 Wenchuan earthquake,” Physica A: Statistical Mechanics and its Applications, vol. 390, no. 12, pp. 2375–2380, 2011. View at Publisher · View at Google Scholar · View at Scopus
  26. B. Krausz and C. Bauckhage, “Loveparade 2010: Automatic video analysis of a crowd disaster,” Computer Vision and Image Understanding, vol. 116, no. 3, pp. 307–319, 2012. View at Publisher · View at Google Scholar · View at Scopus
  27. D. Helbing and P. Mukerji, “Crowd disasters as systemic failures: analysis of the Love Parade disaster,” EPJ Data Science, vol. 1, no. 1, 2012. View at Publisher · View at Google Scholar
  28. M. D'Orazio, L. Spalazzi, E. Quagliarini, and G. Bernardini, “Agent-based model for earthquake pedestrians' evacuation in urban outdoor scenarios: Behavioural patterns definition and evacuation paths choice,” Safety Science, vol. 62, pp. 450–465, 2014. View at Publisher · View at Google Scholar · View at Scopus
  29. Z. Gu, Z. Liu, N. Shiwakoti, and M. Yang, “Video-based analysis of school students’ emergency evacuation behavior in earthquakes,” International Journal of Disaster Risk Reduction, vol. 18, pp. 1–11, 2016. View at Publisher · View at Google Scholar
  30. G. Bernardini, E. Quagliarini, and M. D’Orazio, “Towards creating a combined database for earthquake pedestrians’ evacuation models,” Safety Science, vol. 82, pp. 77–94, 2016. View at Publisher · View at Google Scholar
  31. H. Gayathri, P. M. Aparna, and A. Verma, “A review of studies on understanding crowd dynamics in the context of crowd safety in mass religious gatherings,” International Journal of Disaster Risk Reduction, vol. 25, pp. 82–91, 2017. View at Publisher · View at Google Scholar · View at Scopus
  32. G. K. Still, “Crowd Disasters, Crowd Saf. Risk Anal,” 2018, http://www.gkstill.com/CV/ExpertWitness/CrowdDisasters.html.
  33. J. J. Fruin, “The causes and prevention of crowd disasters,” in Engineering for Crowd Safety, pp. 1–10, 1993. View at Google Scholar
  34. D. C. Duives, W. Daamen, and S. P. Hoogendoorn, “State-of-the-art crowd motion simulation models,” Transportation Research Part C: Emerging Technologies, vol. 37, pp. 193–209, 2013. View at Publisher · View at Google Scholar · View at Scopus
  35. V. J. Kok, M. K. Lim, and C. S. Chan, “Crowd behavior analysis: A review where physics meets biology,” Neurocomputing, vol. 177, pp. 342–362, 2016. View at Publisher · View at Google Scholar · View at Scopus
  36. T. Vicsek and A. Zafeiris, “Collective motion,” Physics Reports, vol. 517, no. 3, pp. 71–140, 2012. View at Publisher · View at Google Scholar · View at Scopus
  37. R. P. Roess, E. S. Prassas, and W. R. Mcshane, Traffic Engineering, Pearson/Prentice Hall, 2011.
  38. N. Shiwakoti, X. Shi, Z. Ye, and W. Wang, “Empirical study on pedestrian crowd behaviour in right angled junction,” in Proceedings of the Australasian Transport Research Forum, p. 12p, 2015. View at Publisher · View at Google Scholar
  39. X. Chen, M. Treiber, V. Kanagaraj, and H. Li, “Social force models for pedestrian traffic – state of the art,” Transport Reviews, vol. 38, no. 5, pp. 625–653, 2018. View at Publisher · View at Google Scholar
  40. X. Jia, H. Yue, X. Tian, and H. Yin, “Simulation of pedestrian flow with evading and surpassing behavior in a walking passageway,” Simulation, vol. 93, no. 12, pp. 1013–1035, 2017. View at Publisher · View at Google Scholar · View at Scopus
  41. M. Haghani and M. Sarvi, “Following the crowd or avoiding it? Empirical investigation of imitative behaviour in emergency escape of human crowds,” Animal Behaviour, vol. 124, pp. 47–56, 2017. View at Publisher · View at Google Scholar · View at Scopus
  42. L. Crociani, A. Gorrini, K. Nishinari, and S. Bandini, “Social Groups and Pedestrian Crowds: Experiment on Dyads in a Counter Flow Scenario,” in Proceedings of the Proceedings of the 8th International Conference on Pedestrian and Evacuation Dynamics (PED2016), Hefei, China, 2016.
  43. M. Kremer and L. G. Debo, “Herding in a Queue: A Laboratory Experiment,” SSRN Electronic Journal, pp. 12–28, 2013. View at Publisher · View at Google Scholar
  44. G. K. Still, Crowd Dynamics, University of Warwick, 2000. View at Publisher · View at Google Scholar
  45. S. Hoogendoorn and W. Daamen, “Self-organization in pedestrian flow,” in Traffic Granul. Flow ’ 03, pp. 373–382. View at Publisher · View at Google Scholar
  46. D. Helbing, P. Molnár, I. J. Farkas, and K. Bolay, “Self-organizing pedestrian movement,” Environment and Planning B: Planning and Design, vol. 28, no. 3, pp. 361–383, 2001. View at Publisher · View at Google Scholar · View at Scopus
  47. D. Helbing, “Traffic and related self-driven many-particle systems,” Reviews of Modern Physics, vol. 73, no. 4, pp. 1067–1141, 2001. View at Publisher · View at Google Scholar · View at Scopus
  48. D. Helbing, A. Johansson, and H. Z. Al-Abideen, “Dynamics of crowd disasters: An Empirical Study,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 75, no. 4, Article ID 046109, 2007. View at Publisher · View at Google Scholar · View at Scopus
  49. J. J. Fruin, Pedestrian Planning and Design, vol. 77, Elevator World, Inc, 1971. View at Publisher · View at Google Scholar
  50. Transportation Research Board, Highway Capacity Manual, National Research Council, Washington, DC, USA, 2010. View at Publisher · View at Google Scholar
  51. A. Gupta and N. Pundir, “Pedestrian flow characteristics studies: a review,” Transport Reviews, vol. 35, no. 4, pp. 445–465, 2015. View at Publisher · View at Google Scholar · View at Scopus
  52. U. Weidmann, Transporttechnik der Fussgänger, Transporttechnische Eigenschaften des Fussgängerverkehrs (Literturauswertung), 1993. View at Publisher · View at Google Scholar
  53. V. M. Predtechenskii and A. I. Milinskii, Planning for Foot Traffic in Buildings (translated from the Russian), 1969.
  54. R. A. Klein, SFPE Handbook of Fire Protection Engineering, 5th edition, 2016. View at Publisher · View at Google Scholar
  55. W. H. K. Lam, J. F. Morrall, and H. Ho, “Pedestrian flow characteristics in Hong Kong,” Transportation Research Record Journal of the Transportation Research Board, vol. 1995, pp. 56–62, 1995. View at Google Scholar
  56. W. Daamen and S. Hoogendoorn, “Controlled experiments to derive walking behaviour,” European Journal of Transport and Infrastructure Research, vol. 3, pp. 39–59, 2003. View at Google Scholar
  57. A. Seyfried, B. Steffen, W. Klingsch, and M. Boltes, “The fundamental diagram of pedestrian movement revisited,” Journal of Statistical Mechanics: Theory and Experiment, no. 10, Article ID P10002, pp. 41–53, 2005. View at Publisher · View at Google Scholar · View at Scopus
  58. J. Zhang, W. Klingsch, A. Schadschneider, and A. Seyfried, “Transitions in pedestrian fundamental diagrams of straight corridors and T-junctions,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2011, no. 6, Article ID P06004, 2011. View at Publisher · View at Google Scholar · View at Scopus
  59. J. Zhang and A. Seyfried, “Empirical Characteristics of Different Types of Pedestrian Streams,” Procedia Engineering, vol. 62, pp. 655–662, 2013. View at Publisher · View at Google Scholar
  60. C.-J. Jin, R. Jiang, S. C. Wong et al., “Large-scale pedestrian flow experiments under high-density conditions,” 2017, https://arxiv.org/abs/1710.10263.
  61. X. Liu, W. Song, and J. Zhang, “Extraction and quantitative analysis of microscopic evacuation characteristics based on digital image processing,” Physica A: Statistical Mechanics and its Applications, vol. 388, no. 13, pp. 2717–2726, 2009. View at Publisher · View at Google Scholar · View at Scopus
  62. U. Chattaraj, A. Seyfried, and P. Chakroborty, “Comparison of pedestrian fundamental diagram across cultures,” Advances in Complex Systems (ACS), vol. 12, no. 3, pp. 393–405, 2009. View at Publisher · View at Google Scholar · View at Scopus
  63. J. Zhang, A. Tordeux, and A. Seyfried, “Effects of boundary conditions on single-file pedestrian flow,” in Proceedings of the 9th International Conference on Cellular Automata, ACRI 2014, J. Wąs, G. C. Sirakoulis, and S. Bandini, Eds., pp. 462–469, Springer International Publishing, Cham, Switzerland, 2015. View at Publisher · View at Google Scholar
  64. S. Cao, J. Zhang, D. Salden, J. Ma, C. Shi, and R. Zhang, “Pedestrian dynamics in single-file movement of crowd with different age compositions,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 94, no. 1, 2016. View at Publisher · View at Google Scholar
  65. J. Wang, M. Boltes, A. Seyfried, J. Zhang, V. Ziemer, and W. Weng, “Linking pedestrian flow characteristics with stepping locomotion,” Physica A: Statistical Mechanics and its Applications, vol. 500, pp. 106–120, 2018. View at Publisher · View at Google Scholar · View at Scopus
  66. S. Cao, J. Zhang, W. Song, C. Shi, and R. Zhang, “The stepping behavior analysis of pedestrians from different age groups via a single-file experiment,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2018, no. 3, p. 033402, 2018. View at Publisher · View at Google Scholar
  67. M. Boltes, A. Seyfried, B. Steffen, and A. Schadschneider, “Automatic Extraction of Pedestrian Trajectories from Video Recordings,” in Pedestrian and Evacuation Dynamics, W. W. F. Klingsch, C. Rogsch, A. Schadschneider, and M. Schreckenberg, Eds., pp. 43–54, Springer, Berlin, Germany, 2010. View at Publisher · View at Google Scholar
  68. M. S. Sharifi, K. Christensen, A. Chen, D. Stuart, Y. S. Kim, and Y. Chen, “A large-scale controlled experiment on pedestrian walking behavior involving individuals with disabilities,” Travel Behaviour and Society, vol. 8, pp. 14–25, 2017. View at Publisher · View at Google Scholar · View at Scopus
  69. M. S. Sharifi, Analysis and modeling of pedestrian walking behaviors involving individuals with disabilities, Utah State University, 2016, https://digitalcommons.usu.edu/etd/4959.
  70. M. S. Sharifi, D. Stuart, K. Christensen, and A. Chen, “Traffic Flow Characteristics of Heterogeneous Pedestrian Stream Involving Individuals with Disabilities,” Transportation Research Record, vol. 2537, pp. 111–125, 2015. View at Publisher · View at Google Scholar
  71. M. S. Sharifi, D. Stuart, K. Christensen, and A. Chen, “Time Headway Modeling and Capacity Analysis of Pedestrian Facilities Involving Individuals with Disabilities,” Transportation Research Record, vol. 2553, pp. 41–51, 2016. View at Publisher · View at Google Scholar
  72. T. Kretz, A. Grünebohm, A. Kessel, H. Klüpfel, T. Meyer-König, and M. Schreckenberg, “Upstairs walking speed distributions on a long stairway,” Safety Science, vol. 46, no. 1, pp. 72–78, 2008. View at Publisher · View at Google Scholar · View at Scopus
  73. X. L. Zhang, W. G. Weng, H. Y. Yuan, and J. G. Chen, “Empirical study of a unidirectional dense crowd during a real mass event,” Physica A: Statistical Mechanics and its Applications, vol. 392, no. 12, pp. 2781–2791, 2013. View at Publisher · View at Google Scholar · View at MathSciNet
  74. J. L. Berrou, J. Beecham, P. Quaglia, M. A. Kagarlis, and A. Gerodimos, “Calibration and validation of the Legion simulation model using empirical data,” in Pedestrian and Evacuation Dynamics, pp. 167–181, 2007. View at Publisher · View at Google Scholar
  75. A. John, A. Schadschneider, D. Chowdhury, and K. Nishinari, “Trafficlike collective movement of ants on trails: absence of a jammed phase,” Physical Review Letters, vol. 102, no. 10, pp. 1–4, 2009. View at Publisher · View at Google Scholar
  76. S. Wang and W. Song, “Experimental Study of Ant Movement in a Straight Passageway under Stress Conditions,” Journal of Insect Behavior, vol. 29, no. 6, pp. 735–743, 2016. View at Publisher · View at Google Scholar · View at Scopus
  77. C. Dias, O. Ejtemai, M. Sarvi, and M. Burd, “Exploring Pedestrian Walking through Angled Corridors,” Transportation Research Procedia, vol. 2, pp. 19–25, 2014. View at Publisher · View at Google Scholar
  78. C. Jin, R. Jiang, J. Yin, L. Dong, and D. Li, “Simulating bi-directional pedestrian flow in a cellular automaton model considering the body-turning behavior,” Physica A: Statistical Mechanics and its Applications, vol. 482, pp. 666–681, 2017. View at Publisher · View at Google Scholar
  79. C. Dias and R. Lovreglio, “Calibrating cellular automaton models for pedestrians walking through corners,” Physics Letters A, vol. 382, no. 19, pp. 1255–1261, 2018. View at Publisher · View at Google Scholar
  80. N. Shiwakoti, M. Sarvi, G. Rose, and M. Burd, “Consequence of Turning Movements in Pedestrian Crowds during Emergency Egress,” Transportation Research Record, vol. 2234, no. 1, pp. 97–104, 2011. View at Publisher · View at Google Scholar
  81. C. Dias, M. Sarvi, N. Shiwakoti, and M. Burd, “Turning Angle Effect on Emergency Egress: Experimental Evidence and Pedestrian Crowd Simulation,” Transportation Research Record, vol. 2312, no. 1, pp. 120–127, 2012. View at Publisher · View at Google Scholar
  82. A. Matsui, T. Mashiko, and T. Nagatani, “Traffic flow of mobile objects through obstacles: Turning and translational objects,” Physica A: Statistical Mechanics and its Applications, vol. 388, no. 2-3, pp. 157–173, 2009. View at Publisher · View at Google Scholar · View at Scopus
  83. D. Yanagisawa, A. Kimura, A. Tomoeda et al., “Introduction of frictional and turning function for pedestrian outflow with an obstacle,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 80, no. 3, Article ID 036110, 2009. View at Publisher · View at Google Scholar · View at Scopus
  84. J. Zhang, W. Klingsch, T. Rupprecht, A. Schadschneider, and A. Seyfried, Empirical Study of Turning And Merging of Pedestrian Streams in T-Junction, 2011.
  85. N. Shiwakoti, M. Sarvi, C. Dias, and M. Burd, “Understanding crowd panic at turning and intersection through model organisms,” in Pedestrian and Evacuation Dynamics, pp. 1175–1183, Springer International Publishing, Cham, Switzerland, 2014. View at Publisher · View at Google Scholar
  86. S. Burghardt, A. Seyfried, and W. Klingsch, “Performance of stairs—fundamental diagram and topographical measurements,” Transportation Research Part C: Emerging Technologies, vol. 37, pp. 268–278, 2013. View at Publisher · View at Google Scholar · View at Scopus
  87. C. Dias, O. Ejtemai, M. Sarvi, and N. Shiwakoti, “Pedestrian Walking Characteristics through Angled Corridors,” Transportation Research Record, vol. 2421, no. 1, pp. 41–50, 2014. View at Publisher · View at Google Scholar
  88. C. Dias, M. Sarvi, O. Ejtemai, and M. Burd, “Elevated Desired Speed and Change in Desired Direction,” Transportation Research Record, vol. 2490, pp. 65–75, 2015. View at Publisher · View at Google Scholar
  89. C. Dias, M. Sarvi, and M. Asano, “Optimal trajectories for constrained pedestrian turning manoeuvres,” in Proceedings of the 8th International Conference on Pedestrian and Evacuation Dynamics, pp. 1–9, Hefei, China, 2016.
  90. M. S. Sharifi, D. Stuart, K. Christensen, A. Chen, Y. S. Kim, and Y. Chen, “Analysis of walking speeds involving individuals with disabilities in different indoor walking environments,” Journal of Urban Planning and Development, vol. 142, no. 1, 2014. View at Publisher · View at Google Scholar
  91. A. Sieben, J. Schumann, A. Seyfried, and D. R. Chialvo, “Collective phenomena in crowds—Where pedestrian dynamics need social psychology,” PLoS ONE, vol. 12, no. 6, p. e0177328, 2017. View at Publisher · View at Google Scholar
  92. L. Sun, L. Cui, S. Qiu, L. Yao, and J. Rong, “Impact on pedestrian flow of bends in passenger access tunnels,” Proceedings of the Institution of Civil Engineers - Transport, pp. 1–10, 2017. View at Publisher · View at Google Scholar
  93. A. Jelić, C. Appert-Rolland, S. Lemercier, and J. Pettré, “Properties of pedestrians walking in line: Fundamental diagrams,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 85, no. 3, 2012. View at Publisher · View at Google Scholar
  94. M. Moussaïd, E. G. Guillot, M. Moreau et al., “Traffic instabilities in self-organized pedestrian crowds,” PLoS Computational Biology, vol. 8, no. 3, Article ID e1002442, 2012. View at Publisher · View at Google Scholar · View at Scopus
  95. H. Kuang, M.-J. Cai, X.-L. Li, and T. Song, “Asymmetric effect on single-file dense pedestrian flow,” International Journal of Modern Physics C, vol. 26, no. 6, 1550064, 13 pages, 2015. View at Publisher · View at Google Scholar · View at MathSciNet
  96. N. A. Rahman, N. A. Alias, and S. H. M. Adenan, “Empirical investigation of trajectories and desired walking velocity of pedestrian walking through angled-corridor,” in AIP Conference Proceedings, AIP Publishing LLC, 2017. View at Publisher · View at Google Scholar
  97. C. Dias, M. Sarvi, N. Shiwakoti, O. Ejtemai, and M. Burd, “Investigating collective escape behaviours in complex situations,” Safety Science, vol. 60, pp. 87–94, 2013. View at Publisher · View at Google Scholar · View at Scopus
  98. R. Nagai, M. Fukamachi, and T. Nagatani, “Evacuation of crawlers and walkers from corridor through an exit,” Physica A: Statistical Mechanics and its Applications, vol. 367, pp. 449–460, 2006. View at Publisher · View at Google Scholar · View at Scopus
  99. J. Zhang, W. Song, and X. Xu, “Experiment and multi-grid modeling of evacuation from a classroom,” Physica A: Statistical Mechanics and its Applications, vol. 387, no. 23, pp. 5901–5909, 2008. View at Publisher · View at Google Scholar · View at Scopus
  100. T. Kretz, A. Grünebohm, and M. Schreckenberg, “Experimental study of pedestrian flow through a bottleneck,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2006, no. 10, pp. P10014–P10014, 2006. View at Publisher · View at Google Scholar
  101. J. Liddle, A. Seyfried, B. Steffen et al., “Microscopic insights into pedestrian motion through a bottleneck, resolving spatial and temporal variations,” 2011, https://arxiv.org/abs/1105.1532.
  102. J. Liddle, A. Seyfried, and B. Steffen, “Analysis of bottleneck motion using Voronoi diagrams,” in Pedestrian and Evacuation Dynamics, pp. 833–836, 2010. View at Publisher · View at Google Scholar
  103. W. Song, J. Zhang, and A. Seyfried, “Experimental Study of Pedestrian Flow in the Channel through Bottleneck,” in Pedestrian and Evacuation Dynamics, R. D. Peacock, E. D. Kuligowski, and J. D. Averill, Eds., pp. 875–879, Springer, Boston, MA, USA, 2011. View at Publisher · View at Google Scholar
  104. W. Tian, W. Song, J. Ma, Z. Fang, A. Seyfried, and J. Liddle, “Experimental study of pedestrian behaviors in a corridor based on digital image processing,” Fire Safety Journal, vol. 47, pp. 8–15, 2012. View at Publisher · View at Google Scholar · View at Scopus
  105. W. Liao, A. Seyfried, J. Zhang, M. Boltes, X. Zheng, and Y. Zhao, “Experimental Study on Pedestrian Flow through Wide Bottleneck,” Transportation Research Procedia, vol. 2, pp. 26–33, 2014. View at Publisher · View at Google Scholar
  106. L. Sun, W. Luo, L. Yao, S. Qiu, and J. Rong, “A comparative study of funnel shape bottlenecks in subway stations,” Transportation Research Part A Policy & Practice, vol. 98, pp. 14–27, 2017. View at Publisher · View at Google Scholar
  107. W. Daamen and S. Hoogendoorn, “Capacity of doors during evacuation conditions,” Procedia Engineering, vol. 3, pp. 53–66, 2010. View at Publisher · View at Google Scholar
  108. W. Daamen and S. P. Hoogendoorn, “Emergency Door Capacity: Influence of Door Width, Population Composition and Stress Level,” Fire Technology, vol. 48, no. 1, pp. 55–71, 2012. View at Publisher · View at Google Scholar · View at Scopus
  109. A. Garcimartín, I. Zuriguel, J. Pastor, C. Martín-Gómez, and D. Parisi, “Experimental Evidence of the “Faster Is Slower” Effect,” Transportation Research Procedia, vol. 2, pp. 760–767, 2014. View at Publisher · View at Google Scholar
  110. A. Garcimartín, D. R. Parisi, J. M. Pastor, C. Martín-Gómez, and I. Zuriguel, “Flow of pedestrians through narrow doors with different competitiveness,” Journal of Statistical Mechanics: Theory and Experiment, no. 4, pp. 1–16, 2016. View at Publisher · View at Google Scholar
  111. N. W. Bode, S. Holl, W. Mehner, A. Seyfried, and G. Xiao, “Disentangling the Impact of Social Groups on Response Times and Movement Dynamics in Evacuations,” PLoS ONE, vol. 10, no. 3, p. e0121227, 2015. View at Publisher · View at Google Scholar
  112. C. von Krüchten and A. Schadschneider, “Empirical study on social groups in pedestrian evacuation dynamics,” Physica A: Statistical Mechanics and its Applications, vol. 475, pp. 129–141, 2017. View at Publisher · View at Google Scholar · View at Scopus
  113. C. von Krüchten, F. Müller, A. Svachiy, O. Wohak, and A. Schadschneider, “Empirical study of the influence of social groups in evacuation scenarios,” in Traffic and Granular Flow '15, pp. 65–72, Springer International Publishing, Cham, Switzerland. View at Publisher · View at Google Scholar
  114. A. Nicolas, S. Bouzat, and M. Kuperman, “Influence of selfish and polite behaviours on a pedestrian evacuation through a narrow exit: A quantitative characterisation,” in Proceedings of the 8th International Conference on Pedestrian and Evacuation Dynamics, Hefei, China, 2016.
  115. A. Nicolas, S. Bouzat, and M. N. Kuperman, “Pedestrian flows through a narrow doorway: Effect of individual behaviours on the global flow and microscopic dynamics,” Transportation Research Part B: Methodological, vol. 99, pp. 30–43, 2017. View at Publisher · View at Google Scholar · View at Scopus
  116. T. Ezaki, K. Ohtsuka, M. Chraibi et al., “Inflow Process of Pedestrians to a Confined Space,” Collective Dynamics, vol. 1, 2016. View at Publisher · View at Google Scholar
  117. X. Liu, W. Song, L. Fu, and Z. Fang, “Experimental study of pedestrian inflow in a room with a separate entrance and exit,” Physica A: Statistical Mechanics and its Applications, vol. 442, pp. 224–238, 2016. View at Publisher · View at Google Scholar · View at Scopus
  118. Legion Software | Science in Motion, Legion Software, 2018, http://www.legion.com/legion-software.
  119. E. Altshuler, O. Ramos, Y. Núñez, J. Fernández, A. J. Batista-Leyva, and C. Noda, “Symmetry breaking in escaping ants,” The American Naturalist, vol. 166, no. 6, pp. 643–649, 2005. View at Publisher · View at Google Scholar · View at Scopus
  120. N. Shiwakoti, M. Sarvi, G. Rose, and M. Burd, “Enhancing the Safety of Pedestrians during Emergency Egress,” Transportation Research Record, vol. 2137, no. 1, pp. 31–37, 2009. View at Publisher · View at Google Scholar
  121. S. Wang, W. Lv, W. Song, and D. R. Chialvo, “Behavior of Ants Escaping from a Single-Exit Room,” PLoS ONE, vol. 10, no. 6, p. e0131784, 2015. View at Publisher · View at Google Scholar
  122. S. Wang, S. Cao, Q. Wang, L. Lian, and W. Song, “Effect of exit locations on ants escaping a two-exit room stressed with repellent,” Physica A: Statistical Mechanics and its Applications, vol. 457, pp. 239–254, 2016. View at Publisher · View at Google Scholar · View at Scopus
  123. N. Shiwakoti and M. Sarvi, “Enhancing the panic escape of crowd through architectural design,” Transportation Research Part C: Emerging Technologies, vol. 37, pp. 260–267, 2013. View at Publisher · View at Google Scholar · View at Scopus
  124. S. Soria, R. Josens, and D. Parisi, “Experimental evidence of the “Faster is Slower” effect in the evacuation of ants,” Safety Science, vol. 50, no. 7, pp. 1584–1588, 2012. View at Publisher · View at Google Scholar
  125. S. Boari, R. Josens, D. R. Parisi, and J. A. Marshall, “Efficient Egress of Escaping Ants Stressed with Temperature,” PLoS ONE, vol. 8, no. 11, p. e81082, 2013. View at Publisher · View at Google Scholar
  126. J. M. Pastor, A. Garcimartín, P. A. Gago et al., “Experimental proof of faster-is-slower in systems of frictional particles flowing through constrictions,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 92, no. 6, 2015. View at Publisher · View at Google Scholar
  127. N. Shiwakoti, M. Sarvi, G. Rose, and M. Burd, “Biologically Inspired Modeling Approach for Collective Pedestrian Dynamics under Emergency Conditions,” Transportation Research Record, vol. 2196, no. 1, pp. 176–184, 2010. View at Publisher · View at Google Scholar
  128. D. R. Parisi, S. A. Soria, and R. Josens, “Faster-is-slower effect in escaping ants revisited: Ants do not behave like humans,” Safety Science, vol. 72, pp. 274–282, 2015. View at Publisher · View at Google Scholar · View at Scopus
  129. A. Sobhani, M. Sarvi, D. Duives, O. Ejtemai, K. Aghabayk, and S. Hoogendoorn, “Exploring the relationship of exit flow and jam density in panic scenarios using animal dynamics,” Transportation Research Procedia, pp. 745–751, 2014. View at Publisher · View at Google Scholar
  130. A. Sobhani, M. Sarvi, D. Duives, O. Ejtemai, K. Aghabayk, and S. Hoogendoorn, “Exploring the Relationship of Exit Flow and Jam Density in Panic Scenarios Using Animal Dynamics,” Transportation Research Procedia, vol. 2, pp. 745–751, 2015. View at Publisher · View at Google Scholar
  131. C. Saloma, G. J. Perez, G. Tapang, M. Lim, and C. Palmes-Saloma, “Self-organized queuing and scale-free behavior in real escape panic,” Proceedings of the National Acadamy of Sciences of the United States of America, vol. 100, no. 21, pp. 11947–11952, 2003. View at Publisher · View at Google Scholar · View at Scopus
  132. C. Saloma, G. J. Perez, C. A. Gavile, J. J. Ick-Joson, and C. Palmes-Saloma, “Prior individual training and self-organized queuing during group emergency escape of mice from water pool,” PLoS ONE, vol. 10, no. 2, 2015. View at Google Scholar · View at Scopus
  133. P. Lin, J. Ma, T. Liu, T. Ran, Y. Si, and T. Li, “An experimental study of the “faster-is-slower” effect using mice under panic,” Physica A: Statistical Mechanics and its Applications, vol. 452, pp. 157–166, 2016. View at Publisher · View at Google Scholar
  134. P. Lin, J. Ma, T. Y. Liu et al., “An experimental study of the impact of an obstacle on the escape efficiency by using mice under high competition,” Physica A: Statistical Mechanics and its Applications, vol. 482, pp. 228–242, 2017. View at Publisher · View at Google Scholar · View at Scopus
  135. Y. Zhang, J. Ma, Y. Si et al., “Required width of exit to avoid the faster-is-slower effect in highly competitive evacuation,” Chinese Physics B, vol. 26, no. 8, p. 084504, 2017. View at Publisher · View at Google Scholar
  136. J. M. Chen, P. Lin, F. Y. Wu, D. L. Gao, and G. Y. Wang, “Revisit the faster-is-slower effect for an exit at a corner,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2018, no. 2, p. 023404, 2018. View at Publisher · View at Google Scholar
  137. H. Oh and J. Park, “Main factor causing “faster-is-slower” phenomenon during evacuation: rodent experiment and simulation,” Scientific Reports, vol. 7, no. 1, 2017. View at Publisher · View at Google Scholar
  138. I. Zuriguel, D. R. Parisi, R. C. Hidalgo et al., “Clogging transition of many-particle systems flowing through bottlenecks,” Scientific Reports, vol. 4, no. 1, 2014. View at Publisher · View at Google Scholar
  139. A. Garcimartín, J. M. Pastor, L. M. Ferrer, J. J. Ramos, C. Martín-Gómez, and I. Zuriguel, “Flow and clogging of a sheep herd passing through a bottleneck,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 91, no. 2, 2015. View at Publisher · View at Google Scholar
  140. I. Zuriguel, J. Olivares, J. M. Pastor et al., “Effect of obstacle position in the flow of sheep through a narrow door,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 94, no. 3, 2016. View at Publisher · View at Google Scholar
  141. M. Isobe, T. Adachi, and T. Nagatani, “Experiment and simulation of pedestrian counter flow,” Physica A: Statistical Mechanics and its Applications, vol. 336, no. 3-4, pp. 638–650, 2004. View at Publisher · View at Google Scholar · View at Scopus
  142. R. Nagai, M. Fukamachi, and T. Nagatani, “Experiment and simulation for counterflow of people going on all fours,” Physica A: Statistical Mechanics and its Applications, vol. 358, no. 2-4, pp. 516–528, 2005. View at Publisher · View at Google Scholar · View at Scopus
  143. T. Kretz, A. Grünebohm, M. Kaufman, F. Mazur, and M. Schreckenberg, “Experimental study of pedestrian counterflow in a corridor,” Journal of Statistical Mechanics: Theory and Experiment, no. 10, Article ID P10001, 2006. View at Publisher · View at Google Scholar · View at Scopus
  144. Z. Yang, Y. Zhang, and O. Grembek, “Combining traffic efficiency and traffic safety in countermeasure selection to improve pedestrian safety at two-way stop controlled intersections,” Transportation Research Part A: Policy and Practice, vol. 91, pp. 286–301, 2016. View at Publisher · View at Google Scholar · View at Scopus
  145. M. Muramatsu, T. Irie, and T. Nagatani, “Jamming transition in pedestrian counter flow,” Physica A: Statistical Mechanics and its Applications, vol. 267, no. 3-4, pp. 487–498, 1999. View at Publisher · View at Google Scholar · View at Scopus
  146. Q. Zhang, B. Han, and D. Li, “Modeling and simulation of passenger alighting and boarding movement in Beijing metro stations,” Transportation Research Part C: Emerging Technologies, vol. 16, no. 5, pp. 635–649, 2008. View at Publisher · View at Google Scholar · View at Scopus
  147. S. Seriani and R. Fernandez, “Pedestrian traffic management of boarding and alighting in metro stations,” Transportation Research Part C: Emerging Technologies, vol. 53, pp. 76–92, 2015. View at Publisher · View at Google Scholar · View at Scopus
  148. Y. Tajima and T. Nagatani, “Scaling behavior of crowd flow outside a hall,” Physica A: Statistical Mechanics and its Applications, vol. 292, no. 1-4, pp. 545–554, 2001. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  149. L. Sun, S. Hao, Q. Gong, S. Qiu, and Y. Chen, “Pedestrian roundabout improvement strategy in subway stations,” in Proceedings of the Institution of Civil Engineers - Transport, vol. 171, pp. 20–29, 2017. View at Publisher · View at Google Scholar
  150. D. Versluis, Microscopic Interaction Behavior between Individual Pedestrians, 2010. View at Publisher · View at Google Scholar
  151. J. Zhang, W. Klingsch, A. Schadschneider, and A. Seyfried, “Ordering in bidirectional pedestrian flows and its influence on the fundamental diagram,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2012, Article ID P02002, 2012. View at Publisher · View at Google Scholar · View at Scopus
  152. S. Xue, R. Jiang, B. Jia, Z. Wang, and X. Zhang, “Pedestrian counter flow in discrete space and time: experiment and its implication for CA modelling,” Transportmetrica B: Transport Dynamics, pp. 1–16, 2017. View at Publisher · View at Google Scholar
  153. Q. Wang, W. Song, J. Zhang, and S. Lo, “Bi-directional movement characteristics of Camponotus japonicus ants during nest relocation,” Journal of Experimental Biology, 2018. View at Publisher · View at Google Scholar
  154. M. Plaue, M. Chen, G. Bärwolff, and H. Schwandt, “Trajectory Extraction and Density Analysis of Intersecting Pedestrian Flows from Video Recordings,” in Photogrammetric Image Analysis, vol. 6952 of Lecture Notes in Computer Science, pp. 285–296, Springer, Berlin, Germany, 2011. View at Publisher · View at Google Scholar
  155. S. C. Wong, W. L. Leung, S. H. Chan et al., “Bidirectional pedestrian stream model with oblique intersecting angle,” Journal of Transportation Engineering, vol. 136, no. 3, pp. 234–242, 2010. View at Publisher · View at Google Scholar · View at Scopus
  156. J. Wu and S. Lu, “Feature Analysis and Operation Evaluation of Pedestrian Weaving Zone,” Transportation Research Record, vol. 2393, no. 1, pp. 66–74, 2013. View at Publisher · View at Google Scholar
  157. L. Lian, X. Mai, W. Song, Y. K. Kit Richard, X. Wei, and J. Ma, “An experimental study on four-directional intersecting pedestrian flows,” Journal of Statistical Mechanics: Theory and Experiment, vol. 2015, Article ID P08024, 2015. View at Publisher · View at Google Scholar
  158. L. Sun, Q. Gong, S. Hao, C. Wang, and Y. Chen, “Experimental Study of Oblique Pedestrian Streams,” PROMET - TrafficTransportation, vol. 30, pp. 151–10, 2018. View at Publisher · View at Google Scholar
  159. K. E. Boyce, D. A. Purser, and T. J. Shields, “Experimental studies to investigate merging behaviour in a staircase,” Fire and Materials, vol. 36, no. 5-6, pp. 383–398, 2012. View at Publisher · View at Google Scholar · View at Scopus
  160. X. Shi and W. Wang, Characteristics of Merging Pedestrian Flows at Congested Situation, 2015.
  161. F. Huo, W. Song, L. Chen, C. Liu, and K. M. Liew, “Experimental study on characteristics of pedestrian evacuation on stairs in a high-rise building,” Safety Science, vol. 86, pp. 165–173, 2016. View at Publisher · View at Google Scholar · View at Scopus
  162. K. Aghabayk, M. Sarvi, O. Ejtemai, and A. Sobhani, “Impacts of Different Angles and Speeds on Behavior of Pedestrian Crowd Merging,” Transportation Research Record, vol. 2490, pp. 76–83, 2015. View at Publisher · View at Google Scholar
  163. K. Aghabayk, O. Ejtemai, M. Sarvi, and A. Sobhani, “Understanding Pedestrian Crowd Merging Behavior,” Transportation Research Procedia, vol. 2, pp. 768–773, 2014. View at Publisher · View at Google Scholar
  164. Z. Shahhoseini, M. Sarvi, and M. Saberi, “Insights toward characteristics of merging streams of pedestrian crowds based on experiments with panicked ants,” Transportation Research Record, vol. 2561, pp. 81–88, 2016. View at Publisher · View at Google Scholar
  165. L. Lian, X. Mai, W. Song, Y. K. K. Richard, Y. Rui, and S. Jin, “Pedestrian merging behavior analysis: An experimental study,” Fire Safety Journal, vol. 91, pp. 918–925, 2017. View at Publisher · View at Google Scholar · View at Scopus
  166. N. Shiwakoti, Y. Gong, X. Shi, and Z. Ye, “Examining influence of merging architectural features on pedestrian crowd movement,” Safety Science, vol. 75, pp. 15–22, 2015. View at Publisher · View at Google Scholar · View at Scopus
  167. Z. Shahhoseini, M. Sarvi, and M. Saberi, “Pedestrian crowd dynamics in merging sections: Revisiting the “faster-is-slower” phenomenon,” Physica A: Statistical Mechanics and its Applications, vol. 491, pp. 101–111, 2018. View at Publisher · View at Google Scholar · View at Scopus
  168. M. Chen, J. Wang, Y. Zhi, and L. Gao, “Effect of the Ratio of the Branch Inflow to the Total Inflow on Evacuation Efficiency of Pedestrians Merging at T-junctions,” Procedia Engineering, vol. 211, pp. 70–77, 2018. View at Publisher · View at Google Scholar
  169. J. Ma, W. Song, W. Tian, S. Lo, and G. Liao, “Experimental study on an ultra high-rise building evacuation in China,” Safety Science, vol. 50, no. 8, pp. 1665–1674, 2012. View at Publisher · View at Google Scholar
  170. Z. Shahhoseini, M. Sarvi, M. Saberi, and M. Haghani, “Pedestrian Crowd Dynamics Observed at Merging Sections,” Transportation Research Record, vol. 2622, pp. 48–57, 2017. View at Publisher · View at Google Scholar
  171. A. Cuesta, O. Abreu, A. Balboa, and D. Alvear, “An experimental data-set on merging flows in rail tunnel evacuation,” Tunnelling and Underground Space Technology, vol. 70, pp. 155–165, 2017. View at Publisher · View at Google Scholar · View at Scopus
  172. M. Haghani, M. Sarvi, Z. Shahhoseini, M. Boltes, and J. Jing, “How Simple Hypothetical-Choice Experiments Can Be Utilized to Learn Humans’ Navigational Escape Decisions in Emergencies,” PLoS ONE, vol. 11, no. 11, p. e0166908, 2016. View at Publisher · View at Google Scholar
  173. M. Haghani and M. Sarvi, “Stated and revealed exit choices of pedestrian crowd evacuees,” Transportation Research Part B: Methodological, vol. 95, pp. 238–259, 2017. View at Publisher · View at Google Scholar · View at Scopus
  174. J. R. G. Dyer, C. C. Ioannou, L. J. Morrell et al., “Consensus decision making in human crowds,” Animal Behaviour, vol. 75, no. 2, pp. 461–470, 2008. View at Publisher · View at Google Scholar · View at Scopus
  175. B. Zhou, X. Wang, and X. Tang, “Understanding collective crowd behaviors: Learning a Mixture model of Dynamic pedestrian-Agents,” in Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2871–2878, Providence, RI, USA, June 2012. View at Publisher · View at Google Scholar
  176. M. Campanella, S. Hoogendoorn, and W. Daamen, “Improving the Nomad microscopic walker model,” IFAC Proceedings Volumes, vol. 42, no. 15, pp. 12–18, 2009. View at Publisher · View at Google Scholar
  177. T. Robin, G. Antonini, M. Bierlaire, and J. Cruz, “Specification, estimation and validation of a pedestrian walking behavior model,” Transportation Research Part B: Methodological, vol. 43, no. 1, pp. 36–56, 2009. View at Publisher · View at Google Scholar · View at Scopus
  178. E. Cepolina and N. Tyler, “Understanding Capacity Drop for designing pedestrian environments, Walk 21 Everyday Walk. Cult,” (2005), http://discovery.ucl.ac.uk/1412/1/Cepolina_Tyler_paper_Walk21.pdf.
  179. M. Asano, T. Iryo, and M. Kuwahara, “Microscopic pedestrian simulation model combined with a tactical model for route choice behaviour,” Transportation Research Part C: Emerging Technologies, vol. 18, no. 6, pp. 842–855, 2010. View at Publisher · View at Google Scholar · View at Scopus
  180. J. R. G. Dyer, A. Johansson, D. Helbing, I. D. Couzin, and J. Krause, “Leadership, consensus decision making and collective behaviour in humans,” Philosophical Transactions of the Royal Society B: Biological Sciences, vol. 364, no. 1518, pp. 781–789, 2009. View at Publisher · View at Google Scholar · View at Scopus
  181. M. Moussaïd, D. Helbing, S. Garnier, A. Johansson, M. Combe, and G. Theraulaz, “Experimental study of the behavioural mechanisms underlying self-organization in human crowds,” Proceedings of the Royal Society B Biological Science, vol. 276, no. 1668, pp. 2755–2762, 2009. View at Publisher · View at Google Scholar · View at Scopus
  182. W. Daamen and S. Hoogendoorn, “Calibration of pedestrian simulation model for emergency doors by pedestrian type,” Transportation Research Record, vol. 2316, pp. 69–75, 2012. View at Publisher · View at Google Scholar · View at Scopus
  183. D. R. Parisi, P. A. Negri, and L. Bruno, “Experimental characterization of collision avoidance in pedestrian dynamics,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, vol. 94, no. 2, 2016. View at Publisher · View at Google Scholar
  184. R. Guo, S. C. Wong, Y. Xia, H. Huang, W. H. Lam, and K. Choi, “Empirical Evidence for the Look-Ahead Behavior of Pedestrians in Bi-directional Flows,” Chinese Physics Letters, vol. 29, no. 6, p. 068901, 2012. View at Publisher · View at Google Scholar
  185. S. Xie, S. C. Wong, W. H. K. Lam, and A. Chen, “Development of a bidirectional pedestrian stream model with an oblique intersecting angle,” Journal of Transportation Engineering, vol. 139, no. 7, pp. 678–685, 2013. View at Publisher · View at Google Scholar · View at Scopus
  186. M. Boltes, J. Zhang, A. Seyfried, and B. Steffen, “T-junction: Experiments, trajectory collection, and analysis,” in Proceedings of the IEEE International Conference on Computer Vision Workshops (ICCV '11), pp. 158–165, Barcelona, Spain, November 2011. View at Publisher · View at Google Scholar
  187. J. Wu, S. Lu, Y. Wang, and X. Liu, “Level of Service Evaluation Method of Pedestrian Weaving Area,” in Proceedings of the 96th Annual Meeting on Transportation Research Board, pp. 1–16, Washington, DC, USA, 2017.
  188. W. Liao, A. Tordeux, A. Seyfried et al., “Measuring the steady state of pedestrian flow in bottleneck experiments,” in Physica A: Statistical Mechanics and its Applications, pp. 1–13, 2016. View at Publisher · View at Google Scholar
  189. L. Sun, Z. Yang, J. Rong, and X. Liu, “Study on the Weaving Behavior of High Density Bidirectional Pedestrian Flow,” Mathematical Problems in Engineering, vol. 2014, Article ID 765659, 9 pages, 2014. View at Publisher · View at Google Scholar
  190. J. Qiao, L. Sun, X. Liu, and J. Rong, “Reducing the impact of speed dispersion on subway corridor flow,” Applied Ergonomics, vol. 65, pp. 362–368, 2017. View at Publisher · View at Google Scholar · View at Scopus
  191. Y. Liu, X. Shi, Z. Ye, N. Shiwakoti, and J. Lin, “Controlled experiments to examine different exit designs on crowd evacuation dynamics,” in CICTP 2016: Green and Multimodal Transportation and Logistics, pp. 779–790, American Society of Civil Engineers, Reston, VA, USA, 2016. View at Publisher · View at Google Scholar
  192. N. Shiwakoti, X. Shi, Z. Ye, Y. Gong, and W. Wang, “Empirical Investigation on merging crowd behaviour with and without blocked vision,” in Proceedings of the World Conference on Transport Research, Shanghai, China, 2016.
  193. Z. Shahhoseini, M. Sarvi, and E. Ito, “Collective movements of pedestrians: How we can learn from simple experiments with non-human (ant) crowds,” PLoS ONE, vol. 12, no. 8, p. e0182913, 2017. View at Publisher · View at Google Scholar
  194. W. Liao, A. U. Kemloh Wagoum, and N. W. Bode, “Route choice in pedestrians: determinants for initial choices and revising decisions,” Journal of the Royal Society Interface, vol. 14, no. 127, p. 20160684, 2017. View at Publisher · View at Google Scholar
  195. G. Warnaby, Marketing Strategy, JMark Management LLC, 1999, http://ezproxy.library.capella.edu/login?url=http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=5659377&site=ehost-live&scope=site.
  196. N. Shiwakoti, “Understanding differences in emergency escape and experimental pedestrian crowd egress through quantitative comparison,” International Journal of Disaster Risk Reduction, vol. 20, pp. 129–137, 2016. View at Publisher · View at Google Scholar · View at Scopus
  197. S. Gwynne, E. Kuligowski, K. Boyce et al., “Enhancing egress drills: Preparation and assessment of evacuee performance,” Fire and Materials. View at Publisher · View at Google Scholar
  198. A. Leeson, P. Alvarez, and S. Ghosh, “Understanding How Big Data and Crowd Movements will Shape the Cities of Tomorrow,” in Proceedings of the European Transport Conference, Frankfurt, Germany, 2014, http://abstracts.aetransport.org/paper/index/id/4296/confid/19.
  199. X. Shi, X. Zhao, Z. Ye, N. Shiwakoti, and J. Lu, “Estimating pedestrian walking characteristics by use of smartphone sensing: an experimental study,” in Proceedings of the 17th COTA conference International Conference of Transportation Professionals (CICTP2017), American Society of Civil Engineers, 2017. View at Publisher · View at Google Scholar
  200. R. Lovreglio, V. Gonzalez, and Z. Feng, “Prototyping virtual reality serious games for building earthquake preparedness: the auckland city hospital case study,” 2018, https://arxiv.org/abs/1802.09119.
  201. Civil-Security-and-Traffic, Jülich-Supercomuting-Center, and Forschungszentrum-Jülich, Data Archive of Experimental Data from Studies about Pedestrian Dynamics, 2018, http://ped.fz-juelich.de/db/.
  202. Pablo Augusto Negri, GSdataset-PANKit - Traffic Intersection Video Sequence DataSet, 2017, http://pablonegri.free.fr/Downloads/GSdataset-PANKit.htm.
  203. CSAI, Vittorio Emanuele II Gallery Dataset, An Annot. Video about Pedestr. Gr. Dyn. Proxemic Behav, 2013.
  204. B. Zhou, Grand Central Station Dataset, Train Stn. Dataset, 2012, http://www.ee.cuhk.edu.hk/~xgwang/grandcentral.html.
  205. B. Majecka, Edinburgh Informatics Forum Pedestrian Database, 2010, http://homepages.inf.ed.ac.uk/rbf/FORUMTRACKING/.
  206. B. Zhou, Collective Motion Database, 2013, http://mmlab.ie.cuhk.edu.hk/projects/collectiveness/dataset.htm.
  207. W. Song and L. Fu, “Verification and validation methods,” in Evacuation Model, pp. 81–102, 2016. View at Publisher · View at Google Scholar