Research Article  Open Access
Research on the Effects of Heterogeneity on Pedestrian Dynamics in Walkway of Subway Station
Abstract
The major objective of this paper is to study the effects of heterogeneity on pedestrian dynamics in walkway of subway station. We analyze the observed data of the selected facility and find that walking speed and occupied space were varied in the population. In reality, pedestrians are heterogeneous individuals with different attributes. However, the research on how the heterogeneity affects the pedestrian dynamics in facilities of subway stations is insufficient. The improved floor field model is therefore presented to explore the effects of heterogeneity. Pedestrians are classified into pedestrians walking in pairs, fast pedestrians, and ordinary pedestrians. For convenience, they are denoted as Ppedestrians, Fpedestrians, and Opedestrians, respectively. The proposed model is validated under homogeneous and heterogeneous conditions. Three pedestrian compositions are simulated to analyze the effects of heterogeneity on pedestrian dynamics. The results show that Ppedestrians have negative effect and Fpedestrians have positive effect. All of the results in this paper indicate that the capacity of walkway is not a constant value. It changes with different component proportions of heterogeneous pedestrians. The heterogeneity of pedestrian has an important influence on the pedestrian dynamics in the walkway of the subway station.
1. Introduction
Pedestrian dynamics plays an important role in the study of pedestrian flow, crowd evacuations, and so on. The research of the pedestrian dynamics in walkways focuses on the selforganization phenomenon [1–3] and jamming transition [4–9]. A variety of simulation models have been developed to study the pedestrian dynamics, such as social force model [1, 10–12], cellular automaton (CA) model [13–17], lattice gas model [18–20], and multiagent model [21, 22]. CA model has been widely used because of its simple rules and good performance for reproducing various selforganization phenomena. In [3, 23–25], a more sophisticated CA model—floor field model—has been introduced, which by now has become the standard CA approach to pedestrian dynamics [26]. In this model, there are two kinds of floor field, that is, static and dynamic ones. The static floor field is constant in time and represents the constant properties of the infrastructure. The dynamic floor field describes the dynamic interactions between the pedestrians. Furthermore, it has its own dynamics, namely, diffusion and decay, which leads to dilution and finally the vanishing of the trace after some time [23].
To determine the adequacy of the proposed model, it is essential to execute the process of validation. Validation has two types, qualitative and quantitative. Tsiftsis et al. [27] proposed a cellularautomatabased model that estimates the movement of individuals. The efficiency of the model has been thoroughly validated with qualitative major characteristics of crowd behavior and quantitative flowdensity dependence. Georgoudas et al. [28] made use of empirical and simulation results to clarify the operation of the anticipative crowd management system and evaluate its efficiency. Klüpfel and MeyerKönig [29] validated the presented model through comparing the flowdensity fundamental diagram of the simulation result with the empirical one and calibrated six parameters. Wąs et al. [30] proposed a new discrete model and validated the model with fundamental diagram. Qu et al. [31] validated the presented microscopic spatialcontinuous model with walking speed, flow characteristics, lane forming behavior, and fundamental diagram. Campanella et al. [32] proposed a simple validation procedure that combines qualitative and quantitative assessments.
Most of these existing models treated pedestrian crowd as a collection of the same isolated individuals, that is, homogeneous pedestrians, which does not coincide with the reality. In reality, the pedestrian flow is composed of heterogeneous people with different individual attributes, including gender, age, luggage, and walking together or not. Their attributes reflect different microscopic traffic characteristics involving occupied space, speed, stride, stride frequency, and so forth. At present, some scholars have put their efforts to research the effects of heterogeneity on pedestrian flow. Lu et al. [33] proposed an extended floor field CA model to incorporate into the walking behavior of pedestrian groups and found the walking behavior of groups has an important but negative influence on pedestrian flow dynamics, especially when the density is at a high level. Campanella et al. [34] chose desired speed, body size, and reaction time as the heterogeneity parameters. Their simulation results strongly indicate that the impact of heterogeneity is very important and should not be neglected in modeling and analyzing pedestrian flow operations. Matsumoto et al. [35] analyzed the heterogeneous pedestrians of a high demand pedestrian crossing in downtown Tokyo. They have found that pedestrians, depending on their desired speeds, are scanning a certain area to choose their directions and that the changes of direction are getting smaller with increasing speed.
Till now, the effects of heterogeneity in walking facilities of subway stations have not been analyzed. And the floor field CA model has not included the pedestrian heterogeneity completely. Therefore, the major objective of this paper is to research the effects of heterogeneity on pedestrian dynamics by improving the floor field model. The rest of this paper is organized as follows. Section 2 presents the results of the field data and analyzes the characteristics of heterogeneous pedestrian flow. The details of the improved floor field model are introduced in Section 3. The simulation scenario is given in Section 4. Section 5 provides the simulation results to analyze the effects of heterogeneity on pedestrian dynamics. At last, the conclusions are summarized and the direction for future work is illustrated.
2. Data Collection and Characteristics of Heterogeneous Pedestrian Flow
A walkway of Beijing Xizhimen subway station was selected as the observation site. The width and the length of this facility are 3.6 m and 10.4 m (Figure 1), respectively. Two HD cameras were used to record the pedestrian flow for an hour during the peak time of a workday simultaneously. In our observations, 404 pedestrians (213 males, 191 females) were collected in the walkway. Most of the pedestrians were young and middleaged people, and their ages mostly ranged from 15 to 60 years. The proportion of children and the elderly was very low. In our observations, 19% of the pedestrians walk in pairs, 6% of the pedestrians carry large luggage, and more than onethird of the people walk fast.
The scatter plot of flowdensity based on the observed data in the walkway was drawn in Figure 2. The cubic equation is used to fit the curve and the result is good because the goodness of fit is 0.985. The capacity of this facility can be calculated through the flowdensity model [36, 37]. The flow continually grows with the density when it is less than 2.2 pedestrians/m^{2}. When the density reaches 2.2 pedestrians/m^{2}, the flow of the walkway gets the maximum value and the capacity is 1.45 pedestrians/m·s. Then, the flow begins to decrease with the increasing density. The increasing gradient of pedestrian flow is greater than the decreasing one.
We extracted the heterogeneous attributes of pedestrians from the video and organized them in the form of individual cases afterward. Each pedestrian was recorded by individual characteristics, including gender, age, luggage, walking time, and walking in pairs or not. According to the statistical analysis of the observed data, different individual attributes of pedestrians reflect different microscopic traffic characteristics, involving speed, stride, stride frequency, occupied space, and so forth. We select walking speed and occupied space as indicators to represent the pedestrian heterogeneity. All of the observed pedestrians are classified into four classes according to the heterogeneous indexes. The categorization of pedestrians is shown in Table 1. The data listed in Table 1 come from the statistical analysis of the observation video.

Pedestrians who walk in pairs and those with large luggage have the same walking speed and they occupy almost the same spaces. These two classes of pedestrians can be regarded as one class and be collectively called pedestrians walking in pairs. For convenience, pedestrians walking in pairs, fast pedestrians, and ordinary pedestrians are denoted as Ppedestrians, Fpedestrians, and Opedestrians in the rest of this paper.
3. Model Development
3.1. Floor Field Model (FF Model) Introduction
The floor field model (FF model) [23–25] is defined on a twodimensional square lattice where each cell can be occupied by at most one particle (pedestrian). In every time step, each particle is allowed to stay at its current position or move to one of the neighboring cells according to certain transition probabilities (1) given that the destination cell is not occupied by another particle. This is done synchronously for all particles (parallel update). The number of nearest neighbors on the lattice can be either four (von Neumann neighborhood) or eight (Moore neighborhood) [3]: with occupation number ; if the cell is occupied by other particles, the value is 1; otherwise, it is 0. Obstacle number is ; if the cell is occupied by any obstacle, the value is 0; otherwise, it is 1.
is a normalization factor for ensuring : .
The probabilities are given by the interaction with two floor fields, static floor field and dynamic floor field . For the evacuation process, the static floor field describes the shortest distance to the exit. The static floor field does not evolve with time and is not changed by the presence of the pedestrians. Such a field can be used to specify regions of spaces which are more attractive, for example, an emergency exit or shop windows [24]. The dynamic floor field was inspired by the motion of ants, which leave a pheromone trace. Other ants are able to smell this trail and follow it. This concept is adopted in this model. Each particle which moves from one cell to another leaves a virtual trace; that is, the value of the origin cell increases by 1. Therefore, has only nonnegative integer values. This trace acts attractively for other particles due to larger transition probabilities according to (1). The effect is that particles have a tendency to follow each other [38]. The has its own dynamics, namely, diffusion and decay controlled by two parameters and , which means broadening and dilution of the trace. and are two positive parameters for scaling and , respectively. is the coupling to the static floor field which characterizes the knowledge of the shortest path to the exits or the tendency to minimize the costs due to deviation from a planned route. This considerably controls one’s velocity and evacuation times. is the coupling to the dynamic floor field which characterizes the tendency to follow other pedestrians (herding behavior) [39].
3.2. Heterogeneous Floor Field Model (HFF Model)
The space is discretized into cells. In each time step, pedestrians move only one cell in the forward, left, or right directions or remain unmoved, and the backward direction is forbidden for pedestrians [33, 40, 41] in our HFF model (Figure 3). These two fields and determine the transition probabilities in such a way that a particle movement is more likely in the direction of higher fields.
3.2.1. Static Floor Field
In case of the walkway considered here, the static floor field describes the shortest distance to the exit of the walkway. The values of studied here are then calculated with a Manhattan distance metric. The explicit calculating process is as follows:(1)The walkway is divided into a rectangular grid. All of the cells on the right side of the walkway shown in Figure 4 are the exit cells (except for the wall cells).(2)Any cell which has the largest distance to the nearest exit cells is assigned a value 0, that is, the cells which are located on the leftmost layer of the walkway (except for the wall cells) in Figure 4.(3)Then, all adjacent cells to the previous one (a “second layer” of cells) are assigned the same value, according to the following rules: if a cell has value , then adjacent cells are assigned the value .(4)Then, the third layer of cells is calculated in the same way as the second layer.(5)The process is repeated until all cells are evaluated.(6)The value of wall cells is . It is beneficial to distinguish wall cells from other cells.
Figure 4 shows the static floor field obtained by applying this set of rules to the walkway of size with exit cells on the right side, where and correspond to the width and the length of the walkway.
3.2.2. Dynamic Floor Field
The dynamic floor field has its own dynamics, namely, decaying with probability and diffusing with probability . After each time step, it is updated according to whereis a discretization of the Laplace operator. Equation (2) can also be obtained by discretizingwhich is the diffusion equation with diffusion constant and an extra term for the decay. Here, is the usual Laplace operator in two dimensions [3].
3.2.3. Update Rules
The update rules of HFF model have the following structure.
(1) Ppedestrians, Fpedestrians, and Opedestrians are generated in the system with random positions. Ppedestrians are distributed in pairs and each group of them move together during the simulation process.
(2) The static floor field of each cell is calculated according to the rules shown in Section 3.2.1.
(3) The dynamic floor field of each cell decays with probability and diffuses with probability to one of its neighboring cells.
(4) For each one of the Fpedestrians and Opedestrians, the transition probabilities for a move to an unoccupied neighbor cell are determined by the local dynamics and the two floor fields. The values of the fields and are weighted with two sensitivity parameters and . This yields
Normalization is as follows:
Here represents the inertia effect [23] for the direction of one’s motion in the previous time step. In the walkway scenario considered in this paper, if a particle moved in the forward direction in the last time step, then given by , and for other directions, where is the sensitivity parameter.
For Ppedestrians, when they want to move left or right, the calculating method of transition probabilities is the same with the isolated pedestrians. But for the case of moving forward, we should compare the transition probabilities and of the cells in front of each pair of Ppedestrians; then the greater one (Figure 5) is chosen as the transition probability to move forward.
(5) Each pedestrian chooses a target cell based on the transition probabilities determined in the previous step.
(6) The conflicts arising from any two or more pedestrians attempting to move to the same target cell are resolved by a probabilistic method [23]. The pedestrians which are allowed to move execute their step with their own speed, and the losers remain unmoved.
(7) at the origin cell of each moving particle is increased by one: .
4. Simulation Scenario
The size of the simulation scenario is the same with the real walkway, and the width and the length are . We discretized the scenario into cells each with a size of . The layout of the walkway is shown in Figure 6. There are three types of pedestrians in the system: Ppedestrians, Fpedestrians, and Opedestrians. For simplicity, the speed of Ppedestrians and Opedestrians is approximate to 1.2 m/s, and Fpedestrians’ speed is approximate to 1.6 m/s. Each time step corresponds to 1/12 s. Pedestrians’ position transition frequency in each time step reflects their speed. The fast pedestrians’ position is updated once in three time steps, and the slow pedestrians’ position is updated once in four time steps. Every pair of Ppedestrians occupies two cells while each one of Fpedestrians and Opedestrians occupies one cell. The periodic boundary is adopted to obtain the flows with different densities. Pedestrians are distributed randomly in the system at the initial time step (Figure 6).
5. Simulation Results
5.1. Model Validation
In order to verify the precision of the proposed model, we validated the model under homogeneous and heterogeneous situations. The capacity and the densityflow fundamental diagrams were reproduced to validate the model quantitatively. The corresponding polynomial trend lines of the observed data and the simulation results have been fitted and compared as shown in [27]. The percentage of the crowd varies from 0.1 to 0.99. The comparison between the observed data and the simulation results is shown in Figures 7(a) and 7(b).
(a) Homogeneous situation
(b) Heterogeneous situation
Under the homogeneous situation, pedestrians are regarded as the same individuals with the average speed of 1.3 m/s when validating the model. Each time step corresponds to 4/13 s. In the process of the simulation, the parameters were set as , , , , and . The simulation result shows that the capacity of the walkway is 1.48 pedestrians/m·s when the density is 2.48 pedestrians/m^{2}. The capacity error between the observed data (1.45 pedestrians/m·s) and the homogeneous simulation result is 2.07%.
There are Ppedestrians, Fpedestrians, and Opedestrians in the system under the heterogeneous situation. The component percentage is 25%, 35%, and 40%. In the process of the simulation, the parameters were set as , , , , and . The capacity of the simulation is 1.43 pedestrians/m·s when the density is 2.43 pedestrians/m^{2}. The capacity error between the observed data (1.45 pedestrians/m·s) and the heterogeneous simulation result is 1.38%.
Moreover, the corresponding polynomial trend lines of observed data and the simulation results of homogeneous and heterogeneous situations have been fitted in Figures 7(a) and 7(b), respectively. High values of goodness of fit () prove that trend lines fit the data sets quite well. The corresponding trend lines share common fundamental properties, regarding degree and end behavior. The capacity values and the fitting curves of the fundamental diagrams coincide well with the observed data, which support the idea that the proposed model is effective and feasible.
5.2. The Effects of PPedestrians on Pedestrian Dynamics
We analyze the effects of Ppedestrians on the pedestrian dynamics through research on the characteristics of the pedestrian flow which is formed with Ppedestrians and Opedestrians. They have the same speed of 1.2 m/s. Each pair of the Ppedestrians occupy two cells and they move together during the simulation process. The percentage of the pedestrian flow increases from 0.1 to 0.99 gradually. The quantitative relationship of flowdensity fundamental diagrams and the curve of capacity with different proportions of Ppedestrians can be obtained (Figures 8 and 9(a)).
(a) Different proportions of Ppedestrians
(b) Different proportions of Fpedestrians
It is found out that Ppedestrians have great impact on the fundamental diagrams. In Figure 8, the differences among these curves are negligible when the density is less than 1.2 pedestrians/m^{2}. Then, when the density is more than 1.2 pedestrians/m^{2}, the gaps among them become large. The location of the fundamental diagrams descends and the capacity of the walkway declines gradually with the increasing percentage of Ppedestrians. The critical density corresponding to the capacity is moving left along the axis with the proportion of Ppedestrians increasing at the same time. It indicates that the phase transition occurs at the lower density with more Ppedestrians in the system.
We extract the capacity values from the fundamental diagrams and plot the curve shown in Figure 9(a). The capacity of the walkway is in the interval of . It always declines proportionally with the increasing percentage of Ppedestrians because the slope of the curve is nearly linear. The results indicate that Ppedestrians have negative effect on walkway capacity. The presence of this kind of pedestrians occupies more spaces and disturbs Opedestrians during the walking process. It may retard the emergence of lane formation and makes the pedestrian flow unsteadily.
5.3. The Effects of FPedestrians on Pedestrian Dynamics
The pedestrian flow is composed of Fpedestrians and Opedestrians in this simulation experiment. Their speeds are 1.6 m/s and 1.2 m/s, respectively. Each of them occupies one cell. The percentage of the pedestrian flow increases from 0.1 to 0.99 gradually. The quantitative relationship of flowdensity fundamental diagrams and the curve of capacity values with different proportions of Fpedestrians can be obtained (see Figures 10 and 9(b)).
With the increasing percentage of Fpedestrians, the changed trend of the curves can be divided into three stages in Figure 10. In the first stage, the curves rise one by one when the proportion of Fpedestrians is less than 0.6. Then, three curves which are corresponding to the proportion values of 0.6, 0.7, and 0.8 almost overlap together. In the third stage, the two curves of 0.9 and 1.0 start to decline when the density is more than 2.2 pedestrians/m^{2} and the 1.0 curve declines faster.
The plot of the capacity whose range is is drawn in Figure 9(b). The capacity grows fast when the proportion is between 0 and 0.3. Then, the increasing speed of the curve becomes slow to almost close to 0 when the proportion is changing from 0.3 to 0.8. The curve begins to go down from the point corresponding to the proportion of 0.8. We can conclude that Fpedestrians have positive impact on the walkway capacity from Figures 9(b) and 10. The mobility of the walkway improves because of the increasing average speed of pedestrians. These two classes of pedestrians cooperate during the walking process when the proportion of Fpedestrians is low. But when the ratio becomes larger and larger, more and more Fpedestrians want to exceed Opedestrians and they change their paths frequently which may change the regime of the pedestrian flow. Pedestrians start to compete when they want to go through the same site. This results in the capacity keeping unchanged and even decreasing.
5.4. The Effects of Different Pedestrian Composition on Pedestrian Dynamics
The pedestrian flow would not have only one type of people because of the random composition of pedestrian crowd in reality. The real pedestrian flow is formed of pedestrians with heterogeneous characteristics. To research the effects of different pedestrian composition on pedestrian dynamics, it needs to set the value of parameters , , and to simulate the pedestrian flow under every combination. According to the condition that , , , and . We set parameters at 0.1 intervals and simulate different parameter combination. The results are shown in Figures 11 and 12.
The result of the experiment shows that the capacity of the walkway ranges from 0.97 to 1.72, and the capacity of pedestrian flow with optimal composition is 77.3% higher than that of pedestrian flow with worst composition. The highest capacity is occupied when the pedestrian flow consists of Fpedestrians and Opedestrians with the ratio being 7 to 3. And we get the lowest capacity when there exist only Ppedestrians. In Figure 11, the curves of fundamental diagrams move up and the capacity increases with less Ppedestrians and more Fpedestrians in pedestrian flow. The combination of , is the proportion of the observed data. The result indicates again that Fpedestrians have the positive effect on the capacity and Ppedestrians have the opposite effect.
We analyze the contour map of the capacity of pedestrian flow in detail. In Figure 12, when the proportion of Ppedestrians ranges from 0% to 50%, the capacity is between 1.06 and 1.72 and when the percentage ranges from 80% to 100%, the capacity is between 0.97 and 1.17. The results indicate that the features of pedestrian flow are related to the proportion of pedestrians with different heterogeneous characteristics. When the proportion of Ppedestrians in the pedestrian flow is large, Fpedestrians will have less influence. However, if the percentage of Ppedestrians is in a comparatively small range (less than 50%), Fpedestrians will have a big influence. Another conclusion is that about onethird of the results is larger than the capacity stipulated in the Code for Design of Metro (the capacity of the unidirectional walkway is 1.39 pedestrians/m·s) of China. It is necessary to take measures to prevent the critical situation caused by some specific composition of heterogeneous pedestrians.
6. Conclusions
In this paper, heterogeneous floor field (HFF) model is proposed to research the pedestrian dynamics of a walkway in Xizhimen subway station. By introducing the pedestrian heterogeneity and analyzing the observed data, we classify the pedestrians into three classes, that is, Ppedestrians, Fpedestrians, and Opedestrians, according to the heterogeneous characteristics of walking speed and space occupied. We validate the precision of the proposed model under the homogeneous and heterogeneous conditions. In order to explore the effects of different heterogeneous characteristics on the pedestrian dynamics, we compared the critical density, fundamental diagrams, and the capacity of pedestrian flow caused by different composition of pedestrians. The existence of the Ppedestrians has negative effect on the pedestrian flow. With the increasing proportion of Ppedestrians, the trend of the fundamental diagrams and the capacity decline continually. By contrast, the Fpedestrians have the positive influence on the whole. The capacity improves with the increasing proportion of Fpedestrians. But the capacity begins to decrease when the percentage of Fpedestrians is larger than 0.8. It reflects that pedestrians start to compete for the spaces and the pedestrian flow is unstable with high ratio of Fpedestrians. In addition, in order to explore the effects of different combination of heterogeneous parameters that may occur in reality on the pedestrian flow, we made the simulation experiment. The results show that Fpedestrians have a big influence when the percentage of Ppedestrians is less than 50%, and when the proportion of Ppedestrians is larger than 80% Fpedestrians have less influence. Another noteworthy result is that about onethird of the capacity is larger than the value stipulated in the Code for Design of Metro of China.
All of the results in this paper indicate that the capacity of walkway is not a constant value. It changes with different component proportions of heterogeneous pedestrians. The heterogeneity of pedestrian has an important influence on the pedestrian dynamics in the walkway of the subway station. These researches can help us understand the macroscopic features of pedestrian flow. They are also beneficial for the operators to take precautions for the emergency caused by special composition of heterogeneous pedestrians in the built stations. And for the unbuilt stations, we can forecast the component proportion of heterogeneous pedestrians in the planning stage to improve the design and construction. In our future work, the presented model will be extended and applied to research the pedestrian dynamics of other facilities in subway stations.
Competing Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
This paper is supported by National Basic Research Program of China (no. 2012CB725403).
References
 D. Helbing, P. Molnár, I. J. Farkas, and K. Bolay, “Selforganizing pedestrian movement,” Environment and Planning B: Planning and Design, vol. 28, no. 3, pp. 361–383, 2001. View at: Publisher Site  Google Scholar
 A. Schadschneider, A. Kirchner, and K. Nishinari, “CA approach to collective phenomena in pedestrian dynamics,” in Cellular Automata, vol. 2493 of Lecture Notes in Computer Science, pp. 239–248, Springer, New York, NY, USA, 2002. View at: Publisher Site  Google Scholar
 S. Nowak and A. Schadschneider, “Quantitative analysis of pedestrian counterflow in a cellular automaton model,” Physical Review E, vol. 85, no. 6, Article ID 066128, 2012. View at: Publisher Site  Google Scholar
 M. Muramatsu, T. Irie, and T. Nagatani, “Jamming transition in pedestrian counter flow,” Physica A: Statistical Mechanics and Its Applications, vol. 267, no. 3, pp. 487–498, 1999. View at: Publisher Site  Google Scholar
 Y. Tajima, K. Takimoto, and T. Nagatani, “Pattern formation and jamming transition in pedestrian counter flow,” Physica A: Statistical Mechanics and Its Applications, vol. 313, no. 34, pp. 709–723, 2002. View at: Publisher Site  Google Scholar
 K. Takimoto, Y. Tajima, and T. Nagatani, “Effect of partition line on jamming transition in pedestrian counter flow,” Physica A: Statistical Mechanics and Its Applications, vol. 308, no. 1–4, pp. 460–470, 2002. View at: Publisher Site  Google Scholar
 H. Kuang, T. Song, X.L. Li, and S.Q. Dai, “Subconscious effect on pedestrian counter flow,” Chinese Physics Letters, vol. 25, no. 4, p. 1498, 2008. View at: Google Scholar
 T. Nagatani, “Freezing transition in bidirectional CA model for facing pedestrian traffic,” Physics Letters, Section A: General, Atomic and Solid State Physics, vol. 373, no. 33, pp. 2917–2921, 2009. View at: Publisher Site  Google Scholar  Zentralblatt MATH
 Y. Tajima, K. Takimoto, and T. Nagatani, “Scaling of pedestrian channel flow with a bottleneck,” Physica A: Statistical Mechanics and its Applications, vol. 294, no. 12, pp. 257–268, 2001. View at: Publisher Site  Google Scholar
 D. Helbing and P. Molnár, “Social force model for pedestrian dynamics,” Physical Review E, vol. 51, no. 5, pp. 4282–4286, 1995. View at: Publisher Site  Google Scholar
 D. Helbing, L. Buzna, A. Johansson, and T. Werner, “Selforganized pedestrian crowd dynamics: experiments, simulations, and design solutions,” Transportation Science, vol. 39, no. 1, pp. 1–24, 2005. View at: Publisher Site  Google Scholar
 D. Helbing, A. Johansson, and H. Z. AlAbideen, “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 Site  Google Scholar
 V. J. Blue and J. L. Adler, “Emergent fundamental pedestrian flows from cellular automata microsimulation,” Transportation Research Record, no. 1644, pp. 29–36, 1998. View at: Google Scholar
 V. J. Blue and J. L. Adler, “Cellular automata microsimulation of bidirectional pedestrian flows,” Transportation Research Record, no. 1678, pp. 135–141, 1999. View at: Google Scholar
 V. J. Blue and J. L. Adler, “Cellular automata microsimulation for modeling bidirectional pedestrian walkways,” Transportation Research Part B: Methodological, vol. 35, no. 3, pp. 293–312, 2001. View at: Publisher Site  Google Scholar
 M. Fukui and Y. Ishibashi, “Jamming transition in cellular automaton models for pedestrians on passageway,” Journal of the Physical Society of Japan, vol. 68, no. 11, pp. 3738–3739, 1999. View at: Publisher Site  Google Scholar
 M. Fukui and Y. Ishibashi, “Selforganized phase transitions in cellular automaton models for pedestrians,” Journal of the Physical Society of Japan, vol. 68, no. 8, pp. 2861–2863, 1999. View at: Publisher Site  Google Scholar
 B. Chopard and M. Droz, Cellular Automata Modeling of Physical Systems, vol. 6, Cambridge University Press, Cambridge, Mass, USA, 2005.
 D. H. Rothman and S. Zaleski, “Latticegas models of phase separation: interfaces, phase transitions, and multiphase flow,” Reviews of Modern Physics, vol. 66, no. 4, pp. 1417–1479, 1994. View at: Publisher Site  Google Scholar
 H.H. Tian, H.D. He, Y.F. Wei, X. Yu, and W.Z. Lu, “Lattice hydrodynamic model with bidirectional pedestrian flow,” Physica A: Statistical Mechanics and Its Applications, vol. 388, no. 14, pp. 2895–2902, 2009. View at: Publisher Site  Google Scholar
 S. Bandini, S. Manzoni, and G. Vizzari, “Agent based modeling and simulation,” in Encyclopedia of Complexity and Systems Science, pp. 184–197, Springer, New York, NY, USA, 2009. View at: Google Scholar
 M. Batty, B. Jiang, and M. ThurstainGoodwin, Local Movement: Agentbased Models of Pedestrian Flows, 1998.
 C. Burstedde, K. Klauck, A. Schadschneider, and J. Zittartz, “Simulation of pedestrian dynamics using a twodimensional cellular automaton,” Physica A: Statistical Mechanics and Its Applications, vol. 295, no. 34, pp. 507–525, 2001. View at: Publisher Site  Google Scholar
 A. Kirchner and A. Schadschneider, “Simulation of evacuation processes using a bionicsinspired cellular automaton model for pedestrian dynamics,” Physica A: Statistical Mechanics and Its Applications, vol. 312, no. 12, pp. 260–276, 2002. View at: Publisher Site  Google Scholar
 A. Kirchner, K. Nishinari, and A. Schadschneider, “Friction effects and clogging in a cellular automaton model for pedestrian dynamics,” Physical Review E, vol. 67, no. 5, Article ID 056122, 2003. View at: Google Scholar
 A. Schadschneider, D. Chowdhury, and K. Nishinari, Stochastic Transport in Complex Systems: From Molecules to Vehicles, Elsevier, New York, NY, USA, 2010.
 A. Tsiftsis, I. G. Georgoudas, and G. C. Sirakoulis, “Real data evaluation of a crowd supervising system for stadium evacuation and its hardware implementation,” IEEE Systems Journal, vol. 10, no. 2, pp. 649–660, 2015. View at: Publisher Site  Google Scholar
 I. G. Georgoudas, G. C. Sirakoulis, and I. T. Andreadis, “An anticipative crowd management system preventing clogging in exits during pedestrian evacuation processes,” IEEE Systems Journal, vol. 5, no. 1, pp. 129–141, 2011. View at: Publisher Site  Google Scholar
 H. Klüpfel and T. MeyerKönig, “Simulation of the evacuation of a football stadium using the CA model PedGo,” in Proceedings of the 5th International Workshop on Traffic and Granular Flow (TGF '03), pp. 423–428, October 2003. View at: Google Scholar
 J. Wąs, R. Lubaś, and W. Myśliwiec, “Proxemics in discrete simulation of evacuation,” in Cellular Automata, pp. 768–775, Springer, Berlin, Germany, 2012. View at: Google Scholar
 Y.C. Qu, Z.Y. Gao, Y. Xiao, and X.G. Li, “Modeling the pedestrian's movement and simulating evacuation dynamics on stairs,” Safety Science, vol. 70, pp. 189–201, 2014. View at: Publisher Site  Google Scholar
 M. Campanella, S. Hoogendoorn, and W. Daamen, “Quantitative and qualitative validation procedure for general use of pedestrian models,” in Pedestrian and Evacuation Dynamics 2012, pp. 891–905, Springer, 2014. View at: Google Scholar
 L.L. Lu, G. Ren, W. Wang, and Y. Wang, “Modeling walking behavior of pedestrian groups with floor field cellular automaton approach,” Chinese Physics B, vol. 23, no. 8, Article ID 088901, 2014. View at: Publisher Site  Google Scholar
 M. Campanella, S. P. Hoogendoorn, and W. Daamen, “Effects of heterogeneity on selforganized pedestrian flows,” Transportation Research Record, no. 2124, pp. 148–156, 2009. View at: Publisher Site  Google Scholar
 M. Matsumoto, M. Miska, M. Kuwahara, and S. Tanaka, “Microscopic pedestrian simulation considering heterogeneity,” in Proceedings of the 12th World Conference on Transport Research (WCTR '10), Lisbon, Portugal, July 2010. View at: Google Scholar
 A. Schadschneider, H. Klüpfel, T. Kretz, C. Rogsch, and A. Seyfried, “Fundamentals of pedestrian and evacuation dynamics,” in MultiAgent Systems for Traffic and Transportation Engineering, pp. 124–154, 2009. View at: Google Scholar
 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 Site  Google Scholar
 A. Schadschneider, A. Kirchner, and K. Nishinari, “From ant trails to pedestrian dynamics,” Applied Bionics and Biomechanics, vol. 1, no. 1, pp. 11–19, 2003. View at: Publisher Site  Google Scholar
 Y. Zheng, B. Jia, X.G. Li, and N. Zhu, “Evacuation dynamics with fire spreading based on cellular automaton,” Physica A: Statistical Mechanics and Its Applications, vol. 390, no. 1819, pp. 3147–3156, 2011. View at: Publisher Site  Google Scholar
 D.F. Xie, Z.Y. Gao, X.M. Zhao, and D.Z. Wang, “Cellular automaton modeling of the interaction between vehicles and pedestrians at signalized crosswalk,” Journal of Transportation Engineering, vol. 138, no. 12, pp. 1442–1452, 2012. View at: Publisher Site  Google Scholar
 I. G. Georgoudas, G. Koltsidas, G. C. Sirakoulis, and I. T. Andreadis, “A cellular automaton model for crowd evacuation and its autodefined obstacle avoidance attribute,” in Cellular Automata, pp. 455–464, Springer, Berlin, Germany, 2010. View at: Google Scholar
Copyright
Copyright © 2016 Haoling Wu 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.