Introduction
In the serious earthquake that hit East Japan on March 11, 2011, 92.4% of the fatalities were caused by drowning, which is totally different from other previous earthquakes where casualties were mainly caused by Tsunami. Since this disaster, a great deal of attention has been directed to the topic of evacuation from earthquake and Tsunami (Wegscheider et al., 2011; Charnkol et. al., 2006; Limanond et al., 2011) and researchers attempted to calculate the time that takes to reach shelters and the evacuation time of citizens (Lämmel et al., 2010; Proulx, 1995; Radwan et al., 1985). Different evacuation behavior may be seen under serious disaster situations (Helbing et al., 2000; Sugimoto et al., 2003; Charnkol et. al, 2006; Kim et al., 2017; Lee et al., 2018) and it is difficult to observe and collect real data under disaster due to a limitation of detecting the coverage area. This study attempts to identify and classify characteristics of walking behavior in normal situations as well as in emergency situations with the points of view from psychology, biology and cognitive science. This study focuses on the fundamentals of human behavior to explain how people walk under normal conditions and why they encounter problems when they must evacuate from a dangerous area. The decision making theory (Janis et al., 1977) suggests that people’s decisions are based on the utility of each individual. However, it has not been explained that people’s behavior could vary depending on the different situations of danger.
Generally, people tend not to pay so much attention on how and where they walk in normal situations especially on routine routes. However, when a serious fire or earthquake occurs, people try to avoid the dangerous situation as soon as possible. If people try to escape based on the decision making theory in these emergency situations, the casualties could be reduced significantly. Nevertheless, people often hesitate and waste their time, or they do not choose correct escape routes even when they know the location of the nearest shelter. Why do people change their behavior in emergency conditions and fail to make rational decisions? To explain these abnormal behaviors, this study attempts to understand human behavior in more detail with the points of view from psychology and cognitive science.
Literature Review
The walking characteristics of pedestrians have mainly been interpreted in terms of crowd flow. Understanding all the elements of individual pedestrian behavior is a very complex task, however, the general view of researchers is that it is possible to generalize behavioral patterns of pedestrians if viewed as crowd flow. Researchers have aggregated individuals into crowd flows and approached this behavior from the cognitive aspect of individuals, in other words, from the socio-cultural aspect. Also, efforts have been made to give explanation through systems that have similar mechanics, for example, fluid dynamics and queueing in transportation, operation research.
Researches that attempt to understand the pedestrian behavior psychologically tend to take an approach of examining religious events and crowd psychology.
Chakraborty et al.(2009) have approached crowd flow in the socio-cultural aspect. Especially, (Coleman et al., 1961; Cheng et al., 2014; Aveni, 1977; Harmeet et al., 2009) identified the characteristics of pedestrian behavior in religious meetings. In addition, crowd flow has been highlighted by many researchers in the aspect of crowd psychology (Turner, 1991; Zhao et al., 2008; Bellomo et al., 2008; Lorenz et al.,2011; Durupınar et al., 2015).
However, most of the research based on crowd flow has been performed in a way that tried to understand crowd flow through similar systems rather than taking on a sociological approach. From this point of view, the walking characteristics of pedestrians have been studied by substituting pedestrian flow with crowd flow and understanding the characteristics. This approach includes studies that attempt to understand crowd flow based on crowd behavior in emergency and panic situations (Helbing et al., 2001; Isobe et al., 2004; McGuire, 2005).
Similar systems that can be utilized to model crowd flow include the queues used in transportation system as well as fluids and hydrodynamics used in dynamics. Studies tried to understand crowd flow through queues (Helbing et al., 2013; Coleman et al., 1961), possibly because it is suitable to model through queues when the crowd movement is toward a specific target point in the situation where the waiting time is uncertain. A special case of queues like bottleneck caused by irregular geometric conditions of sidewalks have also been studied (Hoogendoorn, et al., 2005; Rupprecht et al., 2011). Aveni(1977) classified crowd movement into three types and compared them each other. The motion of pedestrian crowd was modeled as the motion of gases and fluids, crossing behavior of stationary pedestrian crowd was modeled as pedestrian movement in the form of river-like streams, and footsteps of pedestrians were modeled as streamlines of fluids. Kadanoff(1985) used hydrodynamics and analyzed pedestrians walking in opposite directions using viscous fingering. The attempts to study pedestrian characteristics through observations of similar systems are well documented by Helbing et al.(2001).
Researches that apply the concept of fluid flow consider a crowd as a continuum with a flow. However, as a crowd is a group of objects with the ability to think and is expected to obey rational and scientific flow rules, it would be possible to forecast crowd flows more realistically if they can be compared with other similar forms according to McPhail(1991). Such studies that apply similar systems are also applied in the approach of understanding flocking behavior (Limanond et al., 2001; Moussaïd et al., 2010; Qingge et al., 2007). This is because the flow of flocking behavior is a collection of behaviors that follow the simple rules of individual emergency situations rather than a flow formed by the central control or coordination.
Efforts to replace human walking with crowd flow and understand it using other similar systems have a limitation that the nature of human walking can be obscured by the characteristics of the similar system used. For example, in transportation system, the movement of the queue is modeled by factors such as acceleration and arrival rate of individual vehicles whose behavior is controlled within lanes. However, it is difficult to reflect the irregularities of human walking caused by sudden changes of direction. Therefore, this study seeks to approach human walking through the psychology and perception of individuals that govern their behavior.
The purpose of this study is to focus on understanding the human walking itself in psychological and cognitive aspects. This approach is easy to understand because walking is understood in terms of human behavior, and the selection of the parameters of behavioral characteristics through this can contribute to improving the reality of the modelling of walking behavior. The following sections are organized into three chapters. In chapter 3, we classify human walking conceptually. In chapter 4, we demonstrate the conceptually classified human walking based on actual case data. Finally, chapter 5 concludes with a summary of the study and suggestions for future research.
Characterization of Human Walking
Freud(1923) suggested that intelligence and basic instinct are utilized to explain all decisions of human behavior. This study applies this theory to define the fundamentals of walking behavior of human. Based on Freud’s theory, walking behavior can be divided into two: intellectual and instinct.
A behavior can be called intellectual if it is learned by repetition, practice and study because learning is a more proximate definition to explain the intellectual behavior of human. Walking behavior resulted from purposes to find the shortest path or detour are mainly related to intellectual behavior because making actions based on what you learned can be interpreted as having intentions.
People should be concentrating on their movements when they walk along streets, however, people tend not to. This can be explained by the “Stroop effect” suggested by Stroop et al.(1935) as shown in Fig. 1. If someone asks us about the color of the word in Fig. 1, in which the color of the word ‘WHITE’ is black, we have to think more carefully even if saying the color out loud is easier than silent reading.
Fig.1. Stroop effect
Even though reading involves higher intellectual skills than saying, it has been trained so much and used automatically. ‘Automaticity’ explains a demonstration of reducing the reaction time of a task inside a brain, because when people learn and practice something enough, the brain follows a customized or habitualized manual in their own memory to save its energy. Langer(1989) explained the reduction of reaction time of a task inside a brain in that ‘Automaticity’ has a different way of thinking and is usually the result of learning, repetition, and practice so it is related to learning and practices. For example, people usually don’t think much while walking in a normal routine situation. At first, people have to learn how to walk or think to choose the route and direction and have some rules in mind to avoid collision with others and so on. As time goes by, these actions are repeated and people require less effort to think. Finally, these actions change into automaticity.
On the other hand, if a behavior occurs without studying or practicing, it might be a kind of instinctual behavior naturally taken over from the ancestor. The abnormal running or hesitation that get seen in evacuations from fire accidents can be explained mainly by the instinctual movement.
In a nutshell, three different walking behaviors can be derived. In a normal situation, people mainly use ‘automaticity’ and do not need to concentrate on walking so much in a habitual situation. When the street is crowded, people concentrate on walking and mainly use ‘intellectualization’ to reach the destination in a short time or to avoid collisions with others. However, in a serious emergency situation, evacuee behavior is different because people easily fall into a panic state (Parisi et al., 2005; Helbing et al., 2000) so their behavior may follow ‘instinct’ as depicted in Fig. 2.
Fig. 2. Conceptualization of human walking
Demonstration
This chapter will look into how the walking behavior conceptually classified into three types can be matched with actual data such as CCTV and YouTube videos.
Data
This study collected data of people’s trajectory to define walking behavior under normal conditions and emergency situations. For walking behavior under normal conditions, pedestrian movements at Shinjuku station (Fig. 3(A.a)) were collected on January 26, 2011, with the total of 1,632 people during one hour. For walking behavior under emergency situations, the study collected trajectory data at Narita Airport (Fig. 3(B.a, B.b, B.c)) and videos from YouTube (Fig. 3(C.a, C.b, C.c), Fig. 3(D.a, D.b, D.c)).
Fig. 3. Different sources of data collection for walking behavior
Due to lack of data, it is difficult to observe real human behavior in emergency situations and only a few cases were found on YouTube. The first case is a video of the inside of Narita Airport in the event of an earthquake. In this case, two evacuees suddenly run outside because they felt the shaking as can be seen in Fig. 3(B.a). A few seconds later, more people run outside to follow these two people (Fig. 3(B.b)) and finally, everyone rushes out through the door in a short period of time (Fig. 3(B.c)). Fortunately, this video has a yellow line on the floor on the left side of the frame and the thickness of this line enables the calculation of the speed and walking directions of people, which will be discussed next chapter. In the second case as shown in Fig. 3(C.a, C.b, C.c), three people are in the house when the earthquake occurs. The exact distance and speed of the movements of people cannot be extracted from this video, but the movements can be observed for around two minutes. When the earthquake occurs, people are shocked as can be seen in Fig. 3(C.a), they hesitate for a while (Fig. 3(C.b)), and finally they try to go outside and open the door (Fig. 3(C.c)). In the last case as shown in Fig. 3(D.a, D.b, D.c), some people do not move and hold onto a pole when the earthquake occurs as can be seen in Fig. 3(C.a) and 3(B.a), and others walk together or stand on the street even when the earthquake stops after a few minutes.
Methodology
With the data collected, walking behaviors of people such as speed and change of direction are observed and analyzed with the method shown in Fig. 4, which is to observe the differences between normal walking and busy walking. Equation [1] expresses the average change of direction which is the average of the sum of the absolute values of the angles rotating left and right based on the direction of travel every 0.1 seconds. In other words, it is the sum of the absolute values of the angles between the direction of destination (direction II) and the direction of movement diverted from the destination (direction I) at the time of observation. Equation [2] refers to the average speed measured every 0.1 seconds.
Fig. 4. Method of analyzing walking behavior
Results
With proposed method, non-emergency walking behaviors at Shinjuku station in Fig. 3(A.a) is analyzed. In contrast to Fig. 3(B.a, B.b, B.c), 3(C.a, C.b, C.c), and 3(D.a, D.b, D.c), pedestrian movements in Fig. 3(A.a) are relatively stable in speed and direction, which can be seen in normal situations. In this study, movements with stable speed and direction are defined as normal walking, and movements with varying direction and higher speed, which can be found during peak hours, is defined as busy walking.
In Fig. 5, busy walking is shown as the largest dots on the streets (7.4% of total). The differences in speed and change of direction between normal walking and busy walking are 0.25 m/s and 0.1 degree/s respectively. The difference in change of direction between normal walking and busy walking is very large because busy walking has large variations and the change of directions is more dramatic and sudden compared to normal walking. Average speed of normal walking is 1.23 m/s, which is lower than busy walking.
Fig. 5. Characterization of Non-emergency walking
In terms of speed and change of direction of pedestrians, the movements under emergencies are totally different from normal or busy walking. This is because the behaviors of people mainly followed their instinct at least for a second, so it is necessary to derive a new classification to define this behavior using the instinctual behaviors of animals.
Reflecting this, the study classified walking behaviors into five groups (see Fig. 6). In normal walking (A), people walk along the straight line (average walking speed: 1.23 m/s, average change of direction: 2.8 degree/s in Table 1), but in busy walking (B), they change their speed and direction frequently (variation of speed: 0.11 m/s, variation of change of direction: 67.1 degree/s). In emergency situations, it is assumed that evacuees are in panic and the evacuation behaviors can be divided into three groups: Fast & Straight, Freezing and Tizzy. In the case of Fast & Straight, the average speed (2.0 m/s) is faster than that of busy walking, however, the degree of change of direction is smaller because there is just one direction for escaping and people follow the crowd. The average speed can be very high because evacuees runs as fast as they can and they do not need to change direction to avoid people running towards them.
Fig. 6. Conceptualization of walking behavior
Table 1. Speed and direction of pedestrians and evacuees
*: Not measurable, but expected to be a very small value
**: Not measurable, but expected to be a very large value
We could not measure the speed and direction of “Freezing” and “Tizzy” movements directly from the video, however, we could capture the change of direction which was found to be very large as shown in (D) and (E) in Fig. 6. We combined the patterns in Fig. 6 and the observation results in Table 1 to draw Fig. 7 as the summary of the observation results. The normal walking (A) has smaller variations of speed than busy walking (B), but the speed of Fast & Straight (C) is faster than that of busy walking (B). While Tizzy (E) has the largest value of change of direction among the five behaviors observed, this value of Tizzy (E) is not measured directly so is assumed to be just large. The locations of panic (instinctual) movements from (C) to (E) are located in each corner point in Fig. 7.
Fig. 7. Five types of walking behaviors
Conclusion
This study attempted to analyze human behavior in emergency situations from the aspect of instinct and found that automaticity, or behavior based on learning, is similar to basic instinct. In order to analyze behavior by instinct, this study referred to animal behavior and found that there are differences between behaviors that are based on intellectual learning and those that are not. Busy walking and panic walking showed slight differences in speed and the degree of change of direction, which were also different from those of normal walking. To apply these findings to human behavior, this study collected and analyzed video data of human behavior in normal conditions and emergency situations. As a result, it was found that human behavior in normal conditions comes from automaticity while that in emergency situations comes from basic instinct.
Applying this to animal’s instinct, three behaviors were observed: fast & straight, freezing, and tizzy (hesitating). Under panic conditions, especially, behaviors of freezing and hesitating can be observed from animal behavior and similar behaviors were found from the data of pedestrian evacuation movements.
In conclusion, the method used in this study and the results can be applied to simulations of walking behavior and analysis of behavior in emergency situations. Defining and analyzing pedestrian behavior is not as easy as dealing with vehicle movements. This study can suggest some ways to numerically define pedestrian movement and apply them to designing walking simulation models.
Acknowledgement
This work was supported by the Korea Foundation of Nuclear Safety (KoFONS) [Development of methodology for emergency traffic control in nuclear power plant accident (grant No. 2104040)]
참고문헌
- Aveni, A. (1977). "The not-so-lonely crowd: Friendship groups in collective behavior." Sociometry, Vol. 40, No. 1, pp. 96-99. https://doi:10.2307/3033551.
- Bellomo, N., Christian, D. (2008). "On the modelling crowd dynamics from scaling to hyperbolic macroscopic models." Mathematical Models and Methods in Applied Sciences, Vol. 18, No. supp01, pp. 1317-1345. https://doi.org/10.1142/S0218202508003054
- Chakraborty, P., Roy, G.G., Das, S., Jain, D., Abraham, A. (2009). "An improved harmony search algorithm with differential mutation operator." Fundamenta Informaticae, Vol. 95, No. 4. pp. 401-426. https://doi.org/10.3233/FI-2009-157
- Charnkol, T., Tanaboriboon, Y. (2006). "Tsunami evacuation behavior analysis: One step of transportation disaster response." IATSS Research, Vol. 30, No. 2. pp. 83-96. https://doi.org/10.1016/s0386-1112(14)60173-4
- Cheng, L., Yarlagadda, R., Fookes, C., Yarlagadda, P.K. (2014). "A review of pedestrian group dynamics and methodologies in modelling pedestrian group behaviours." World, Vol. 1, No. 1, pp. 2-13.
- Coleman, J., James, J. (1961). "The equilibrium size distribution of freely-forming groups." Sociometry, Vol. 24, No. 1, pp. 36-45. https://doi.10.2307/2785927.
- Durupinar, F., Gudukbay, U., Aman, A., Badler, N.I. (2015). "Psychological parameters for crowd simulation: From audiences to mobs." IEEE Transactions on Visualization and Computer graphics, Vol. 22, No. 9, pp. 2145-2159. https://doi.org/10.1109/TVCG.2015.2501801
- Freud, S. (1923). The Ego and the Id. https://www.sigmundfreud.net/the-ego-and-the-id-pdf-ebook.jsp.
- Helbing, D., Farkas, I., Vicsek, T. (2000). "Simulating dynamical features of escape panic." Nature, Vol. 407. pp. 487-490. https://doi.org/10.1038/35035023
- Helbing, D., Johansson, A. (2013). "Pedestrian, crowd, and evacuation dynamics." arXiv preprint arXiv:1309.1609.
- Helbing, D., Molnar, P., Farkas, I.J., Bolay, K. (2001). "Self-organizing pedestrian movement." Environment and Planning B: Planning and Design, Vol. 28, No. 3, pp. 361-383. https://doi.org/10.1068/b2697
- Hoogendoorn, S.P., Daamen, W. (2005). "Pedestrian behavior at bottlenecks." Transportation Science, Vol. 39, No. 2, pp. 147-159. https://doi.org/10.1287/trsc.1040.0102
- Isobe M, Helbing D, Nagatani T. (2004). "Experiment, theory, and simulation of the evacuation of a room without visibility." Physical Review E, Vol. 69, No. 6 Pt 2, 066132. https://doi:10.1103/PhysRevE.69.066132.
- Janis, I.L., Mann, L. (1977). "Decision making: A psychological analysis of conflict, choice, and commitment." Free Press, New York.
- Kadanoff, LP. (1985). "From periodic motion to unbounded chaos: Investigations of the simple pendulum." Physica Scripta, Vol. 1985, No. T9, 5. DOI: 10.1088/0031-8949/1985/T9/001.
- Kim, S.-D., Lee, S.-H., Choi, K.-H., Cheung, C.-S. (2017). "A comparative study of on-site action manual for people with disaster vulnerability." Journal of the Society of Disaster Information, Vol. 13, No. 2, pp. 155-162.
- Lammel, G., Rieser, M., Nagel, K., Taubenbock, H., Strunz, G., Goseberg, N., Schlurmann, T., Klupfel, H., Setiadi, N., Birkmann, J. (2010). Emergency Preparedness in the Case of a Tsunami - Evacuation Analysis and Traffic Optimization for the Indonesian City of Padang. Pedestrian and Evacuation Dynamics 2008, Part 1. Springer, Germany, pp. 171-182.
- Langer, E.J. (1989). Minding Matters: The Consequences of Mindlessness-mindfulness. In L. Berkowitz (Ed.). Advances in Experimental Social Psychology, Vol. 22, Academic Press, California, US, pp. 137-173.
- Lee, C.-Y., Kim, T.-H., Lee, C.-S. (2018). "A development of the integrated model for the disaster field response and situation information management." Journal of the Society of Disaster Information, Vol. 14, No. 1, pp. 36-42.
- Limanond, T., Kim, H., Siridhara, S., Tipakornkiat, C., Chermkhunthod, C., Uttra, S. (2011). "Decision on tsunami evacuation route in Tourism area: A case study of had Patong, Phuket." Journal of the Eastern Asia Society for Transportation Studies, Vol. 9, No. 0. pp. 16-30.
- Lorenz, J., Rauhut, H., Schweitzer, F., Helbing, D. (2011). "How social influence can undermine the wisdom of crowd effect." Proceedings of the National Academy of Sciences, Vol. 108, No. 22, pp. 9020-9025. https://doi.org/10.1073/pnas.1008636108
- McGuire, J.M. (2005). "Crowd Management and Evacuation Modeling System." Proceedings of the Southwest Decisions Sciences Institute, Dallas, Texas, USA.
- McPhail. C., (1991). The Myth of the Madding Crowd (1st edition). Routledge. New York, USA.
- Moussaid, M., Perozo, N., Garnier, S., Helbing, D., Theraulaz, G. (2010). "The walking behaviour of pedestrian social groups and its impact on crowd dynamics. PloS one, Vol. 5, e10047. https://doi.org/10.1371/journal.pone.0010047.
- Parisi, D.R., Dorso, C.O. (2005). "Microscopic dynamics of pedestrian evacuation." Physica A: Statistical Mechanics and its Applications, Vol. 354, No. 15. pp. 606-618. https://doi.org/10.1016/j.physa.2005.02.040
- Proulx, G. (1995). "Evacuation Time and movement in apartment buildings." Fire Safety Journal, Vol. 24, No. 3, pp. 229-246. https://doi.org/10.1016/0379-7112(95)00023-M
- Qingge, J., Can, G. (2007). "Simulating crowd evacuation with a leader-follower model." International Journal of Computer Sciences and Engineering Systems, Vol. 1, No. 4, p. 250.
- Radwan, A.E., Hobeika, A.G., Sivasailam, D. (1985). "A computer simulation model for rural network evacuation under natural disasters." Institute of Transportation Engineers, Vol. 55, No. 9, pp. 25-30.
- Rupprecht T., Klingsch W., Seyfried A. (2011). Influence of Geometry Parameters on Pedestrian Flow Through Bottleneck. In: Peacock R., Kuligowski E., Averill J. (eds), Pedestrian and Evacuation Dynamics. Springer, Massachusetts, US. https://doi.org/10.1007/978-1-4419-9725-8_7
- Singh, H., Arter, R., Dodd, L., Langston, P., Lester, E., Drury, J. (2009). "Modelling subgroup behaviour in crowd dynamics DEM simulation." Applied Mathematical Modelling, Vol. 33, No. 12, pp. 4408-4423. https://doi.org/10.1016/j.apm.2009.03.020
- Stroop, J.R., Ridley, J. (1935). "Studies of interference in serial verbal reactions." Journal of Experimental Psychology, Vol. 18, No. 6, pp. 643-662. https://doi.org/10.1037/h0054651
- Sugimoto, T., Murakami, H., Kozuki, Y., Nishikawa, K. (2003). "A human damage prediction method for tsunami disasters incorporating evacuation activities." Natural Hazards, Vol. 29, No. 3. pp. 587-602. https://doi.org/10.1023/A:1024779724065
- Turner, J.C. (1991). Mapping Social Psychology Series. Social Influence, Thomson Brooks, Cole Publishing Co., Wisconsin, USA.
- Wegscheider, S., Post, J., Zosseder, K., Muck, M., Strunz, G., Riedlinger, T., Muhari, A., Anwar, H.Z. (2011). "Generating tsunami risk knowledge at community level as a base for planning and implementation of risk reduction strategies." Natural Hazards and Earth System Sciences, Vol. 11, pp. 249-258. https://doi.org/10.5194/nhess-11-249-2011
- Zhao, D., Yang, L., Li, J. (2008). "Occupants' behavior of going with the crowd based on cellular automata occupant evacuation model." Physica A: Statistical Mechanics and its Applications, Vol. 387, No. 14, pp. 3708-3718. https://doi.org/10.1016/j.physa.2008.02.042