Aversion of Pedestrians to Face-to-Face Situations Eases Crowding Sho Yajima 1 , Kiwamu Yoshii 2+ , and Yutaka Sumino 1 1 Department of Applied Physics, Graduate School of Science, Tokyo University of Science, Katsushika, Tokyo 125-8585, Japan 2 Earthquake Research Institute, The University of Tokyo, Bunkyo, Tokyo 113-0032, Japan (Received April 1, 2020; accepted May 7, 2020; published online June 19, 2020) We conducted a numerical simulation of a crowd of pedestrians. Each pedestrian, modeled with three circles, has a shape whose long axis is perpendicular to the anteroposterior axis and is designed to move in a xed desired direction, i.e., +x or x. Pedestrians have friction at their surface, soft repulsion, and a back force to resist backward motion. In this study, we newly introduced an active rotation that captures the psychological eect of aversion to face-to-face situations. The numerical simulation revealed that active rotation induces the uidization of the system, leading to the higher ux of pedestrians. To reveal the mechanism of the uidization, we calculated the minimum principal stress for each pedestrian. Furthermore, we visualized force chains observed in the system by connecting the minimum principal stresses. We conrmed that the uidization of pedestrians is due to the fragmentation of the force chains induced by the active rotation. 1. Introduction Active matter is a system composed of many elements that transduce energy into motion locally. Such active matter is a novel type of nonequilibrium system showing rich dynamical patterns. Bird ocks, sh schools, and bacterial colonies are well-known examples of active matter. 14) A crowd of pedestrians is also considered as active matter and forms various patterns; 58) for example, lane formation, avalanche behavior, and clogging. We here make a model of a high-density crowd of pedestrians. It is known that a high-density crowd of pedestrians may lead to fatal accidents. 911) Such a critical situation is due to the formation of force chains within the population, which leads to stress concentration at a small number of pedestrians. Thus, understanding the crowd of pedestrians is required to avoid such a dangerous situation. However, the situation is not experimentally accessible, so numerical simulation is a relevant measure to understand the behavior of pedestrians in a high-density crowd. Many numerical simulations have been conducted to understand the behavior of a high-density crowd of pedestrians. 5,1215) In such numerical simulations, a pedestrian can be modeled as a particle that follows Newtons laws of motion, where each individual walks to a xed destination. A crowd of pedestrians is a type of dry active matter , 3) where the momentum of the system is not conserved as the pedestrians move on a rigid substrate. Furthermore, the essential feature of these models is that the physical contact force plays an important role because of the crowded condition. For such reasons, the behavior of a crowd of pedestrians is described as a type of active granular system. Notably, a pedestrian has anisotropy in its function and shape: it has a frontal part directing its motion and is wider in the direction perpendicular to the anteroposterior axis. 16) In this study, we newly adopt an active rotation induced by a psychological eect. The front direction is the preferred direction of movement. Interestingly, however, people have a psychological aversion to face-to-face situations with others. We adopt this eect via the active rotation. Our numerical simulation reveals that the coupling between the anisotropic shape and the active rotation eases crowded conditions for pedestrians. 2. Simulation Setup As shown in Fig. 1, we consider a rectangular system of L x L y and impose the periodic boundary condition in the x direction. In this study, we model a pedestrian i with a central circle whose radius is i and side circles whose radius is i =2. i is xed for each pedestrian and has a homogeneous distribution around the mean. Each pedestrian has a desired direction e i ¼ ð1; 0Þ, a front direction p i ¼ðcos i ; sin i Þ ( i 2 ½%;%Þ), and a velocity v i . A pedestrian tries to walk in the desired direction at a speed v 0 i , and m i is the mass of the pedestrian. v 0 i and m i also have homogeneous distributions around the mean. The position of the center of the pedestrian is denoted by r i ¼ðx i ;y i Þ. The centers of the side circles lie on the circumference of the central one, with the line joining the center perpendicular to p i . The center of the right circle is, r r i ¼ðx r i ;y r i Þ¼ðx i þ i 2 sin i ;y i i 2 cos i Þ and that of the left circle is r l i ¼ðx l i ;y l i Þ¼ðx i i 2 sin i ;y i þ i 2 cos i Þ. We can dene the velocity of the central circle as v i ¼ _ r i ¼ðv x;i ;v y;i Þ, and those of the left and right circles as v r i ¼ðv r x;i ;v r y;i Þ¼ðv x;i þ i ! i 2 cos i ;v y;i þ i ! i 2 sin i Þ and v l i ¼ ðv l x;i ;v l y;i Þ¼ðv x;i i ! i 2 cos i ;v y;i i ! i 2 sin i Þ, respectively. ! i represents the angular velocity of the rotation. The equation of motion for the ith pedestrian is given by m i dv i dt ¼ m i v 0 i e i v i ( þ X ji f ij þ X W f iW þ F b i ; ð1Þ 3 p e v desired front motion x y L x L y 2 Direction of motion Fig. 1. (Color online) Schematic illustration of our pedestrian model and the initial conditions of numerical simulation. Pedestrians are distributed in a rectangular system of L x L y .A xed ratio of pedestrians R walk in the positive x direction, whereas the others walk in the negative x direction. The desired direction of a pedestrian is denoted by e. Each pedestrian has a shape composed of three circles and a front direction p. The central circle is twice as large as the other ones with the line joining the centers. The centers of the side circles lie on the circumference of the central one, with perpendicular to p. Each pedestrian also has a velocity vector v. The color represents the direction of p. Journal of the Physical Society of Japan 89, 074003 (2020) https://doi.org/10.7566/JPSJ.89.074003 074003-1 © 2020 The Physical Society of Japan J. Phys. Soc. Jpn. Downloaded from journals.jps.jp by 3.236.55.199 on 06/23/20