• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Using agent-based simulation to assess disease prevention measures during pandemics?

    2021-09-28 02:18:44YunheTong童蘊賀ChristopherKingandYanghuiHu胡楊慧
    Chinese Physics B 2021年9期
    關(guān)鍵詞:胡楊

    Yunhe Tong(童蘊賀),Christopher King,?,and Yanghui Hu(胡楊慧)

    1Department of Engineering Mathematics,University of Bristol,Bristol BS8 1UB,UK

    2State Key Laboratory of Fire Science,University of Science and Technology of China,Hefei 230027,China

    Keywords:pedestrian dynamics,occupant exposure,COVID-19,simulation study

    1.Introduction

    With the rapid spread of COVID-19 around the world,the World Health Organization(WHO)declared a pandemic.[1]It has fundamentally altered the daily activities of humans all over the world.Governing bodies recognize that intervening strategies are needed to reduce the spread of COVID-19 and so reduce its impact on human health.[2]These measures include banning mass gatherings,introducing social distancing,where people must remain a fixed distance apart,and one-way systems in spaces with high footfalls.These restrictions could be present for months or years to come,[3]so it is worthwhile to attempt to model and quantify the transmission risk of people when they make necessary trips and expose them to other people.

    Modelling epidemiological transmission is crucial to understand the transmission dynamics of the pandemic and provide insights into the design of alternative interventions.[4]In previous work,the susceptible-infected-removed(SIR)model has been used widely to predict the spread of infectious diseases in humans.[5]In this model,people are categorized by S,I,or R(susceptible,infectious,or removed)states and can move between these states.SIR model and its extended modifications have been used to estimate disease spread at large scales.[6,7]However,these models do not explicitly consider the day-to-day interactions between people at smaller scales,such as in buildings.

    Understanding pedestrian dynamics is essential for studying disease transmission within buildings and has been extensively investigated through statistical modelling and simulation.These models can represent realistic pedestrian behaviors and are generally distinguished into macroscopic and microscopic ones.Macroscopic models generate the bulk properties of groups of pedestrians,such as flows and densities,without considering individuals,[8]and cannot be readily used for studying disease transmission.In contrast,microscopic models consider each pedestrian as a self-driven agent which has simple rules for interacting with both the environment and other agents,and aggregating these behaviors over large numbers of agents can lead to interesting collective phenomena.Cellular automata,[9,10]floor field models,[11]and force-based models[12]are common examples of these kinds of models.

    Exposure risk of pedestrians to most diseases in a confined space largely depends on the local interactions between other,potentially infected,pedestrians.Much empirical research has been done to enhance the understanding of interactions.[13–15]In force-based models,interactions of a pedestrian with both others and the environment can be thought of as attractive or repulsive forces acting on said pedestrian.[16,17]For example,a pedestrian is assumed to keep a certain distance from others(personal space),if someone enters this space,the pedestrian will try to move away to maintain their personal space.[18]Alternatively,there are some cases where pedestrians will stay close to others,for example,traveling with relatives or friends,or following guides.[19]

    While previous research has carefully studied pedestrian dynamics in various contexts,little is known about the influences of pedestrian behaviors on the spread of disease,especially in supermarkets,where everyone must visit to buy food.This is one of the essential activities which puts people at greater risk of contracting disease,due to being around others in a confined space.To try to reduce this risk,supermarkets have taken measures such as limiting the maximum number of occupants,introducing directional floor markings and signs to create one-way movement around the store(oneway system),and asking customers to maintain social distance from others.[20]Recently,the“exposed”model has been developed to assess the exposure risk of individuals in confined spaces during pandemics.[21]This model is designed to be used in tandem with microscopic pedestrian dynamics models to investigate different types of disease transmission.The effect of some preventative measures taken in supermarkets,including restricting the number of customers,keeping social distance,and entering with a shopping cart on the average contact degree of each person,has been investigated through simulation.[22]However,these restrictions may increase the amount of time people spend in supermarkets and thus increase their exposure time.So,it is essential to investigate the risk-efficiency trade-off to find the optimal strategy for the shopper.

    In this contribution,we employ a force-based model to implement preventative measures seen during this pandemic,such as social distancing.We focus on the influence of these measures in supermarkets on the exposure risk of customers.Specifically,we investigate the trade-off between exposure risk and the shopping time to develop the optimal shopping strategy when such protective measures are in place.

    There are two main aspects of innovation in this work.First,we develop a simulator that can illustrate disease transmission and pedestrian movement dynamics.It can generate data to assess a variety of metrics regarding strategies to limit the spread of disease and/or optimize pedestrian flow.The other is to establish a model that can evaluate the impact of different disease-prevention measures on customer shopping efficiency and infection risk simultaneously and determine which are the most suitable,given a user’s set of priorities.

    Compared with previous studies,the method we proposed considers how a manager of a supermarket may optimize their store in order to comply with local restrictions while maximizing store turnover.This work also compares shopping strategies that maybe used by customers and determines which has the maximum reduction in exposure risk.More importantly,our model can allow supermarket managers to adjust the weighting of risk and efficiency according to their needs.In addition,this framework allows for the inclusion of empirical data to calibrate the model.

    In the following,we first present an integrated simulation for shopper movements(Subsection 2.1)and choice behavior in a supermarket(Subsection 2.2).We then describe the exposed model and how we use it to assess a shopper’s exposure risk(Subsection 2.3).Using this framework,a simulation study(Subsection 3.1)is conducted,implementing different restrictions,including social distance and one-way systems,along with different shopping habits(Subsection 3.2).Results from this study are presented,examining the tradeoff between customers’exposure risk and shopping efficiency(Subsection 4)for the different scenarios.Finally,we assess the implications of these results on shopping behavior during the current pandemic(Subsection 5).

    2.Model

    We develop an integrated model to study pedestrian movements and shopping behavior in a supermarket.Specifically,we employ a modified force-based model that can vary the social distance between pedestrians to simulate current restrictions(Subsection 2.1).A simple destination choice model is used to govern the activities that shoppers perform(Subsection 2.2).In addition,the exposed model[21]is used to assess the exposure risk of pedestrians over the course of their shopping trip(Subsection 2.3).

    2.1.Force-based model with social distance

    Assuming that each person can be represented by a circle of given radius,the repulsive interaction between two people,i and j,is given by

    The interaction of person i with the wall W is treated analogously as

    where riis the radius of i,diWis the distance between the center of i to the closest point of the boundary of an obstacle,W.niWis the normalized vector acting perpendicularly to diW,and tiWis the vector acting tangentially to diW(perpendicular to niW).

    We use the minimum distance rij?dijto represent the social distance between people.Birepresents the effective length scale over which the interpersonal repulsive force takes effect,such that increasing Biincreases the effective distance between people.To calculate the value of Binecessary for rij?dijto be a specific social distance between two people,i and j,we use the following relation:

    where|vi(t)|is the magnitude of vi(t).Aiand Biare the same as that in Eq.(2).This assumes that the distances from person i to all other people(that are not j)and to all obstacles are much greater than dij,so that their contributions to fijin Eq.(1)are negligible.

    2.2.Pedestrian shopping behavior

    In this simulator,each agent is assigned a set of activities to perform in the supermarket.Each activity can be performed at only one destination in this work,though this need not be the case in general.Based on this schedule of activities,the probability of choosing which activity to perform/destination to visit next is computed using a multinomial logit discrete choice model[23]specified as follows:

    where Psis the probability of an agent choosing destination s,βdis the desirability parameter,and H is the normalization constant which ensures that the sum of choice probabilities over all destinations is unity.dsis the desirability of s,and is determined by the position of the activity associated with s in the activity schedule of the agent,such that

    where a is the position of the activity in the agent’s current schedule.For example,consider a scenario where an agent can choose from one of four destinations,and each has an activity labeled A–D,sequentially.If an agent has the schedule‘A,B,C,D’,then d1=e?1,d2=e?2,d3=e?3,d4=e?4.In this work,an agent’s next destination depends exclusively on its activity schedule,where the destination most likely to be visited is the one with the next activity.To ensure that agents almost always perform the next activity in their schedule,we setβd=20.

    3.Simulations

    3.1.Simulator description

    This section describes the agent-based simulator used in this study,which was written in Java.The simulator can be used to look at both movement and choice behavior of pedestrians in public spaces and is based on the simulator used in Bode and Codling.[26]There is a video showing the simulator in action in the supplementary materials.

    When agents reach a destination,they spend a short amount of time there,mimicking browsing or purchasing behavior.This shopping time,tsis sampled from a shifted exponential distribution,

    where t is the time measured in simulation time-steps(0.05 s),and t0andλare parameters such that t0=5 s andλ=0.01.t0controls the minimum possible waiting time at a destination andλcontrols the spread of possible waiting times.The values of t0andλwere chosen to produce a distribution of waiting times that are reasonable and intuitive.Once they have spent tstimesteps at the destination,they then select their next destination using the model described in Eq.(5).Their activity schedule is updated to remove the associated activity.They do this for each activity in their schedule until it is empty.At this point,the agent’s next destination is the exit,where they then leave the simulation.

    The simulator takes the following of inputs:

    ?the layout of the environment,

    ?the total number of agents,

    ?the total number of time steps to run the simulation for,

    ?the values of any choice model parameters,

    ?the length of the activity schedule for all agents.

    The different supermarket environments used in this study are shown in Fig.1.There are 12 destinations where each has one unique activity that can be performed(so there are 12 activities with only one place to perform each of them).For this study,agents are added at the entrance and initially prescribed a set of randomly allocated activities to perform.The activities are then sorted in order of increasing distance from the entrance to their associated destination.This is done to prevent agents from doubling back in the one-way system scenarios and represents the real-life tendency for supermarket shoppers to visit places closest to the supermarket entrance first.[27]The total size of the simulated spaces is 20×20 m2and,to replicate the capacity constraints imposed during a pandemic,a maximum of 20 agents are allowed in the spaces at any time,corresponding to an average density of 0.05 agents/m2.When there are fewer than 20 agents in the simulation,the simulator places new agents at the entrance,if no other agents are too close(i.e.,beyond social distance).

    The Aiand Bifrom Eq.(1)are fixed across all agents,with Ai=2000 andτi=0.5 s,as originally specified by Helbing et al.[12]Social distancing is implemented by changing the Biparameter(see Subsection 2.1).Under normal conditions,we use Bi=0.08,[12]and using Eq.(4),Bi≈0.15 for 1 m social distancing,and Bi≈0.4 for 2 m social distancing.However,due to the limitations in the simulator,Bicould not be increased above 0.2,so a social distance of 2 m was not possible,and instead we used Bi≈0.2,corresponding to a social distance of 1.2 m.This is smaller than the recommended social distances[28]but should be sufficient to establish general relationships between social distance with both exposure risk and shopping efficiency.For each scenario,the simulation is run for 1-hour of simulated time(72000 time steps at 0.05 s per time step)and repeated ten times.The average exposure risk and shopping efficiency over the 10 replicates is reported.The simulator outputs the positions and directions of each entity at each time step along with whether the agent has left the simulation.

    Fig.1.Destination layout of the supermarket(a)and three environments used in the simulations:the supermarket where agents can move freely(b),the one-way system with narrow passages(c),and one-way with wide passages(d).Each agent enters the supermarket through the entrance and then begins to visit each destination in the assigned schedule.Once their schedule is complete,each person heads toward the exit and are removed from the simulation.The colored stars represent destinations agents may visit.Each agent is represented as a circle colored according to which destination they are heading towards.The line indicates their current direction,and the length of it represents their current speed.

    3.2.Different scenarios

    Several different strategies/scenarios are investigated,looking at both a person’s exposure to others and how quickly they can complete their shopping trip.The different strategies considered are shown in Table 1.Firstly,there is the prepandemic shopping situation,where agents can move around supermarkets freely and do not need to observe social distancing.This“free-for-all with no social distancing”is our base case to which all other scenarios are compared.To look at the effect of social distancing on exposure and shopping efficiency,a scenario where agents can still move freely but observe social distancing(free-for-all with social distancing)is included.To investigate this further,two additional scenarios are defined,where the social distancing is 1 m or above 1 m,as commonly described in government guidance.[5]

    Customers often adopt two different strategies when shopping,both before and during the pandemic,where some shop frequently(e.g.,maybe every 2–3 days),but only shop for a small amount of time/buy only a few items,and others may shop relatively infrequently(e.g.,maybe once a week),but spend more time in the store/buy more items.To assess whether one of these strategies is better for people who are more concerned with minimizing their exposure or maximizing their shopping efficiency,these two strategies are included.These different shopping strategies are represented using the length of agent’s activity schedule as a proxy.

    In some supermarket chains,one-way systems are in place in an attempt to minimize the exposure risk of customers.[20]To test this idea,one-way versions of the freefor-all supermarket environment are created by placing artificial barriers to force entities to move along a prescribed route from the entrance to exit.One has wide walkways,where agents can pass each other,and the other has narrow walkways where they cannot.As mentioned in Subsection 3.1,to ensure that agents obey the one-way system,their schedules are ordered in terms of the distance of the destination associated with that activity from the entrance.The social distance for the scenarios is above 1 m,and each agent visits 5 different destinations in the scenarios unless otherwise specified.

    Table 1.Summary of the different scenarios.

    3.3.Scenario metrics

    COVID-19 can be transmitted in several ways,such as airborne and physical contact.[24]The EXPOSED model[21]allows for several types of disease transmission.When considering exposure by physical contact(the distance of two agents is smaller than two times of body radius)in this simulator,it is almost impossible for two agents to touch one another.This is due to the values of Bineeded to implement social distancing(see Subsection 2.1)and the small number of agents present at any one time(see Subsection 3.1).We consider exposure to others within some distance(here we use d≥1 m),with an additional face-to-face exposure if agents are facing each other within the social distance,defined by a fixed angle of contact(<120°)between the current directions of agents.Therefore,we assess two kinds of exposure:social distance and face-toface contact disease transmission in this work.

    We use the global assessment of exposure G defined as in the exposed model below:[21]

    where l is the number of people an agent is exposed to during a simulation,N is the maximum number of agents to which an agent is exposed,and rlis the risk degree.In this paper,we assume that the risk degree is equal to the number of people an agent is exposed to,i.e.,r1=1 means that an agent is exposed to 1 agent,r2=2 means that an agent is exposed to 2 agents,and so on.Clis cumulative exposure time-steps to l people.According to the literature,[25]the exposure mass percentage at 0 m is 100%,that at 1 m is about 40%,and that at 2 m is about 30%.In our study,the exposed distance we used is 1 m both for social distance method and face-to-face method.We assume that a homogeneous distribution of air in the environment.So the weights of exposure of two methods are weighted equally.

    The cumulative exposure time Clis defined below:

    The exposure time of each agent i exposed to a given number of agents l is calculated as follows:

    where to=1,tf=72000,i.e.,the total number of simulation time-step.When i is exposed to l agents,β=1,otherwise β=0.

    Though preventative measures can reduce the spread of disease,they can also impact the time spent by people in shops.From the perspective of supermarket managers,the rate of throughput(number of people that enter and leave the store per hour)is a good measure of store turnover.Therefore,we define the shopping efficiency as the total number of distinct agents that appeared over the course of the simulation(i.e.,over the course of 1 hour)and then subtracting 20,the maximum occupancy in the simulations.This assumes that there are 20 agents at the last time-step that have not finished their shopping.To consider the fact that agents with shorter schedules will,on average,finish their shopping quicker than those with longer schedules,we standardize shopping efficiency by multiplying by the length of activity schedules in each scenario.Similarly,since the length of activities in eight scenarios is different,to make our exposure results comparable,we standardize G in Eq.(6)by dividing by the schedule length in each scenario.Thus,we get a global assessment of exposure per activity per hour and shopping efficiency as the total number of activities completed over all agents per hour.

    4.Results

    4.1.Effects of social distancing

    We first establish how social distancing influences exposure risk and shopping efficiency.As shown in Fig.2(a),the total exposure due to agents being within social distance decreases as the social distance between agents increases.The exposure due to face-to-face contact also decreases when social distance is enforced.However,it does not make a big difference when the social distance is 1 m and above 1 m.This implies that,in this case,if a disease is primarily transmitted through face-to-face contact,then only relatively small social distances need to be implemented.Despite having no empirical validation,our results show that keeping social distancing is an effective way to reduce costumers’exposure during a pandemic in supermarkets.

    Fig.2.Average values over ten simulation replicates of the global assessment of exposure through social distance and face to face transmission(a)and shopping efficiency of people(b)with increasing social distance.Error bars represent one standard deviation from the average.

    In terms of shopping efficiency(Fig.2(b)),it is obvious that as social distance increases,the agents’shopping efficiency significantly decreases,dropping about 40%when the social distance is above 1 m.One possible explanation for this is that keeping social distance impedes the natural movements of agents,as agents must wait until there is sufficient distance from others before moving or performing activities.

    Our simulation results confirm the effectiveness of social distancing in reducing contact among customers and reveal the decrease in customers’shopping efficiency when implemented in supermarkets.

    4.2.Effects of shopping strategy

    Figure 3 shows the exposure and shopping efficiency of agents with different shopping strategies.When costumers have a longer shopping schedule,they have less exposure risk through both social distance and face-to-face contact.One possible reason for this might be that agents who have more activities to complete tend to stay longer in the supermarket and thus relatively reduce the maximum possible number of agents that enter the supermarket while they are present due to the occupancy constraint.The risk of exposure from adding more agents in the confined space might be greater than the risk of agents staying in the store for a long time.

    Fig.3.Average values over ten simulation replicates of the global assessment of exposure through social distance and face to face transmission(a)and shopping efficiency of people(b)when they have different shopping strategies,using schedule length as a proxy.Error bars represent one standard deviation from the average.

    In terms of shopping efficiency(Fig.3(b)),agents with a longer schedule are more likely to have higher shopping efficiency.Shopping efficiency is defined as the total number of activities agents perform in a simulation.Compared to agents with a long schedule,the agents with a short schedule can quickly complete the required shopping activities,while spending roughly the same amount of time entering and leaving the store.Therefore,from a general perspective,agents with a short schedule tend to spend a larger proportion of time entering and leaving the supermarket than those with a long schedule,therefore making better use of their time.This finding indicates the shopping strategy of shopping infrequently but buying more items is better than shopping frequently for a few items both in terms of reducing the exposure and improving shopping efficiency.

    4.3.Effects of one-way system

    Figure 4 shows agents’exposure risk and shopping efficiency with narrow/wide or without a one-way system.It is found that when agents can move freely in the supermarket,they have the highest shopping efficiency and highest exposure risk.There is only one obstacle in the supermarket(see Fig.1),so in this situation,agents with random assigned shopping schedules can direct head for the destinations freely,without considering the influence of the one-way system.It can reduce the time each agent spends on reaching the destinations but increase the chance to contact other agents,leading to higher exposure risk.It is also clear that the exposure risk for both wide and narrow one-way systems is similar,so there is little added benefit from narrowing passages such that people cannot pass each other.

    Fig.4.Average values over ten simulation replicates of the global assessment of exposure through social distance and face to face transmission(a)and shopping efficiency of people(b)both with and without a one-way system with either a wide or narrow passage.Free-for-all is a pre-pandemic shopping situation in which people can move around supermarkets freely.Error bars represent one standard deviation from the average.

    In contrast,when the one-way system is implemented in the supermarket,shopping efficiency decreases,particularly when narrow passages are employed.As agent movements are restricted by the one-way system,they are only allowed to move along pre-designed routes,which are unlikely to be the optimal ones in terms of travel time between destinations.On average,each agent’s speed may significantly decrease,due to other agents blocking the way,particularly when narrow passageways are used.In this case,customers must wait until others complete their shopping activities before they can proceed further along the passageway.

    Therefore,we confirm that the one-way system is effective in lowering the exposure risk of agents.In terms of the two kinds of the one-way system,both have similar exposure risks.However,the one-way system with wide passages only slightly decreases shopping efficiency compared to the prepandemic free-for-all scenario.So,the one-way system with wide passages is a good solution for supermarket managers in balancing infection risk while maintaining average throughput.

    4.4.Trade-off between exposure risk and shopping efficiency

    As discussed above,social distancing,a long shopping schedule,and a one-way system can all reduce the exposure of each agent to others.However,these measures reduce the shopping efficiency of agents of varying degrees.Therefore,it is worth developing a strategy that can take agents’exposure risk and shopping efficiency into consideration simultaneously.

    In this work,we introduce a ranking R,which scores each scenario in terms of minimizing exposure risk while maximizing shopping efficiency.R is given by

    Figure 5 shows the ranking strategies according to the importance of exposure.The strategy with the highest R at any given value ofαis the optimal one in terms of balancing exposure risk with shopping efficiency.The wide one-way system is the best strategy,reducing exposure risk while maintaining shopping efficiency for all values ofα.This implies that implementing one-way systems with passageways wide enough for people to move past each other while obeying social distancing is the ideal strategy for supermarket managers,who would need to implement measures for reducing exposure without compromising store turnover.This is followed closely by the narrow one-way system,except forα=0,where freefor-all with no social distancing higher in rank.However,this assumes that all agents strictly obey the rule of the one-way system,and therefore shows the theoretical result if everyone always obeys the one-way system.As for social distancing,it depends on the customer’s motivation,for example,if a customer is more concerned about minimizing the risk of infection,then they should obey social distancing rules,with larger distances providing more safety.Of course,this assumes that everyone is always equally compliant with social distancing.Finally,customers should shop less frequently but buy more when they do,as the long schedule has both a lower exposure risk and higher shopping efficiency during a situation where social distancing restrictions are in place.

    Fig.5.Ranking strategies as the weight of the exposure risk changes.The strategy with the highest ranking for a given exposure risk weighting is the optimal strategy for shoppers.

    5.Discussion

    This work aims to develop the optimal strategy for minimizing the exposure risk of people to disease while maximizing the shopping efficiency of customers in supermarkets during a pandemic.We propose an integrated agent-based simulator which can model both the decision behavior and movement of people.Agents make decisions on where to go next using a discrete choice model based only on their schedule of desired activities.Step-by-step locomotion is described by a force-based model.We implement social distance restrictions between agents by varying a parameter in the force-based model.A recent disease exposure model[21]is used to assess the exposure risk of shoppers in a supermarket environment.Two kinds of disease transmission are considered,based on known transmission routes of COVID-19:[24]social distance and face-to-face contact.

    We use our simulator to investigate several measures taken to slow down the spread of COVID-19 in supermarkets.These measures are social distancing(≥1 m),and one-way systems.However,these measures are likely to make it harder for shoppers to complete their shopping quickly,so the impact of these protective measures on shopping efficiency is also considered.We also investigate how shopping habits of customers impact their exposure risk,comparing the strategies where customers visit supermarkets less often but spend more time shopping,and the other where customers visit frequently but spend less time shopping.We rank these measures in terms of exposure risk and shopping efficiency to find the optimal shopping strategy during a pandemic.

    Our results confirm previous work[22]that the following government and general health guidelines can reduce the exposure risk of people to disease at the cost of the speed and ease of shopping.This work shows that supermarket managers should adopt one-way systems as they lower customers’exposure risk while preserving turnover,if the passageways are wide enough to allow for customers to move past each other while obeying any other social distancing restrictions.Narrowing the passageways of a one-way system to prevent people passing each other does not decrease overall exposure but does decrease shopping efficiency,so one-way systems should be deployed in stores whose layouts allow for wide passageways.One-way systems could be particularly effective in places where the culture and attitude of people improve compliance to restrictions.For the customers interested in minimizing their exposure to disease,reducing the number of shopping trips but buying more is the optimal strategy,and provides a more efficient use of time,compared to short,frequent store visits.This could be particularly useful in places where people are less likely to obey restrictions,as concerned individuals can benefit from this strategy even if others do not follow disease-prevention measures.

    We use simulations of our proposed model to illustrate the consequences of various measures taken by customers and supermarket managers in terms of exposure risk and shopping efficiency.Although we simplify the scenarios and only consider a few possible factors in disease transmission and customer behavior,we believe that it can achieve reasonable and intuitive results.However,further validation in more complex scenarios is still needed.

    Although some of these results are intuitive,they show that even a simple pedestrian dynamic simulator shows great potential to provide decision-makers and business owners with helpful information to predict infection risks and customer turnover.Of course,there are several limitations to this work that could provide several avenues for future research.

    These include potentially unrealistic assumptions about the behavior of shoppers,for example we assume that customers always obey restrictions all of the time,and that people choose their next destination based only on a pre-defined‘schedule’,whereas in reality,many factors could be important,such as distance traveled.Other behaviors could be included,such as impulse stops,which could be particularly disruptive for one-way systems and reduce their efficacy in both reducing exposure risk and shopping efficiency.The simulated environments could also be expanded to include areas such as checkouts and promotional stands,all of which could influence the amount of time spent in the supermarket and the extent of agent interactions.Perhaps using an environment based on the floorplan of a real supermarket would provide a more realistic setting.Validating these results with empirical data from current or previous pandemics would also be useful to do in the future.These are in addition to the limitations mentioned in previous sections.

    Microscopic pedestrian dynamics simulations could provide reliable and accurate quantitative forecasts for disease transmission,average shopping times,and many other metrics of pedestrian behavior,not just for this current pandemic,but any similar events in the future.

    Acknowledgment

    We are very grateful to Dr Nikolai Bode,who helped guide and improve this project through many helpful discussions and ideas.

    猜你喜歡
    胡楊
    胡楊
    龍門陣(2024年2期)2024-04-18 09:37:37
    《大漠胡楊》
    千年胡楊
    金橋(2021年11期)2021-11-20 06:37:30
    胡楊為什么能在沙漠中生存
    大美胡楊
    北廣人物(2020年47期)2020-12-09 06:24:12
    胡楊之戀
    胡楊 美術(shù)作品欣賞
    大眾文藝(2019年17期)2019-09-09 08:13:04
    家風(fēng)伴我成長
    胡楊頌
    “胡楊講堂”開講——大美天山·大美胡楊
    成年美女黄网站色视频大全免费| 男男h啪啪无遮挡| 男的添女的下面高潮视频| 久久久欧美国产精品| 亚洲自偷自拍图片 自拍| 国产精品亚洲av一区麻豆 | 久久免费观看电影| 国产免费又黄又爽又色| 色播在线永久视频| 日韩一卡2卡3卡4卡2021年| 欧美在线一区亚洲| 欧美97在线视频| 麻豆乱淫一区二区| 欧美人与善性xxx| 久久国产亚洲av麻豆专区| 免费在线观看黄色视频的| 十八禁人妻一区二区| 国产精品.久久久| 久久国产精品男人的天堂亚洲| 国产精品久久久久久精品电影小说| av一本久久久久| 色精品久久人妻99蜜桃| 黄片小视频在线播放| 如何舔出高潮| 亚洲欧美一区二区三区久久| www.精华液| 哪个播放器可以免费观看大片| 99精国产麻豆久久婷婷| 亚洲精品一区蜜桃| 侵犯人妻中文字幕一二三四区| 丁香六月欧美| 亚洲国产av新网站| 亚洲av欧美aⅴ国产| 女性生殖器流出的白浆| 免费黄色在线免费观看| 中文字幕av电影在线播放| 久久久久久久久久久免费av| 亚洲专区中文字幕在线 | 丁香六月天网| 亚洲av综合色区一区| 五月天丁香电影| 精品一区二区三区av网在线观看 | 久久久久久人妻| av女优亚洲男人天堂| 亚洲一级一片aⅴ在线观看| 久久久久人妻精品一区果冻| 性高湖久久久久久久久免费观看| 亚洲成人免费av在线播放| √禁漫天堂资源中文www| 视频在线观看一区二区三区| 波野结衣二区三区在线| 亚洲第一青青草原| 青春草视频在线免费观看| 999精品在线视频| 欧美日韩精品网址| 午夜福利视频精品| 亚洲国产欧美在线一区| 亚洲美女视频黄频| 亚洲美女黄色视频免费看| 成年美女黄网站色视频大全免费| 熟女av电影| 韩国精品一区二区三区| 午夜激情av网站| 成人18禁高潮啪啪吃奶动态图| 日日撸夜夜添| 91aial.com中文字幕在线观看| 大陆偷拍与自拍| 欧美黑人欧美精品刺激| 亚洲图色成人| 精品少妇一区二区三区视频日本电影 | 久久精品国产a三级三级三级| 水蜜桃什么品种好| 男女高潮啪啪啪动态图| 亚洲伊人久久精品综合| 国产极品天堂在线| 免费观看av网站的网址| 亚洲精品美女久久久久99蜜臀 | 熟妇人妻不卡中文字幕| 欧美日韩av久久| 无限看片的www在线观看| 亚洲 欧美一区二区三区| 免费久久久久久久精品成人欧美视频| 日日撸夜夜添| 视频区图区小说| av免费观看日本| 老鸭窝网址在线观看| 久久人妻熟女aⅴ| 欧美黑人欧美精品刺激| 日韩欧美一区视频在线观看| av在线app专区| 伊人久久大香线蕉亚洲五| 国产精品国产三级国产专区5o| 无限看片的www在线观看| 人人澡人人妻人| 两个人看的免费小视频| 亚洲综合精品二区| 97精品久久久久久久久久精品| 九九爱精品视频在线观看| 三上悠亚av全集在线观看| 国产亚洲av高清不卡| 免费在线观看视频国产中文字幕亚洲 | 日韩大片免费观看网站| 亚洲天堂av无毛| 亚洲精品久久久久久婷婷小说| 午夜福利网站1000一区二区三区| 黑丝袜美女国产一区| 国产在线免费精品| 国产成人午夜福利电影在线观看| 国产av一区二区精品久久| 国产亚洲精品第一综合不卡| 国产日韩欧美视频二区| 欧美国产精品va在线观看不卡| av国产精品久久久久影院| 国产极品粉嫩免费观看在线| 满18在线观看网站| 免费女性裸体啪啪无遮挡网站| 午夜福利免费观看在线| 精品酒店卫生间| 亚洲国产精品999| 久久久精品免费免费高清| 天天影视国产精品| 女人高潮潮喷娇喘18禁视频| 在线观看www视频免费| 丝袜脚勾引网站| 日韩欧美精品免费久久| 精品国产国语对白av| av卡一久久| 午夜福利视频精品| 嫩草影院入口| 久久精品国产综合久久久| 9191精品国产免费久久| 日韩欧美精品免费久久| 王馨瑶露胸无遮挡在线观看| 日日爽夜夜爽网站| 久久久久久久精品精品| av一本久久久久| 国产人伦9x9x在线观看| 两个人看的免费小视频| 精品一区二区三区av网在线观看 | 午夜福利,免费看| 日韩 欧美 亚洲 中文字幕| 久久国产精品男人的天堂亚洲| 日韩电影二区| 欧美黑人精品巨大| 亚洲第一av免费看| 日韩电影二区| 国产成人午夜福利电影在线观看| 欧美激情极品国产一区二区三区| 91精品三级在线观看| 亚洲精品一二三| 亚洲国产精品999| 搡老岳熟女国产| 亚洲第一av免费看| 欧美97在线视频| 午夜福利一区二区在线看| 免费看不卡的av| 亚洲在久久综合| 国产免费现黄频在线看| 五月开心婷婷网| 久久久欧美国产精品| 黄片小视频在线播放| 亚洲,欧美精品.| 国产一区二区激情短视频 | 精品一区二区免费观看| 人人妻人人澡人人看| 国产成人91sexporn| 色综合欧美亚洲国产小说| 欧美少妇被猛烈插入视频| 黑人欧美特级aaaaaa片| 久久97久久精品| 欧美日韩国产mv在线观看视频| 亚洲欧美清纯卡通| 水蜜桃什么品种好| 黄网站色视频无遮挡免费观看| 黄色 视频免费看| 日日啪夜夜爽| 日本欧美视频一区| 热99久久久久精品小说推荐| 亚洲精品一二三| 久久av网站| 在线精品无人区一区二区三| 久久久国产精品麻豆| 黑丝袜美女国产一区| 免费观看av网站的网址| 人成视频在线观看免费观看| 亚洲精品aⅴ在线观看| 成人手机av| 国产一级毛片在线| 韩国高清视频一区二区三区| 亚洲精品国产区一区二| 亚洲第一av免费看| 成年人免费黄色播放视频| 宅男免费午夜| 国产一区二区激情短视频 | 大码成人一级视频| 精品午夜福利在线看| 多毛熟女@视频| 久久青草综合色| 亚洲成人手机| 99久国产av精品国产电影| 韩国高清视频一区二区三区| 久久久久久人妻| 免费少妇av软件| 尾随美女入室| 久久久国产精品麻豆| 久久精品国产亚洲av高清一级| 亚洲av国产av综合av卡| 在线观看免费午夜福利视频| 十分钟在线观看高清视频www| 久久久国产一区二区| 成人影院久久| 男男h啪啪无遮挡| 亚洲欧美中文字幕日韩二区| 18禁动态无遮挡网站| 国产亚洲av片在线观看秒播厂| 久久久精品免费免费高清| 黄色一级大片看看| av.在线天堂| 久久精品亚洲av国产电影网| 国产成人免费观看mmmm| √禁漫天堂资源中文www| 亚洲 欧美一区二区三区| 五月天丁香电影| 蜜桃国产av成人99| 男女午夜视频在线观看| 久久久欧美国产精品| 老司机影院毛片| 最新的欧美精品一区二区| 国产麻豆69| 制服人妻中文乱码| 好男人视频免费观看在线| 少妇人妻精品综合一区二区| 伊人久久大香线蕉亚洲五| 精品一品国产午夜福利视频| www.av在线官网国产| 国产女主播在线喷水免费视频网站| 日本av免费视频播放| 少妇的丰满在线观看| 精品一区在线观看国产| 久久久久久久精品精品| 超碰成人久久| 精品国产超薄肉色丝袜足j| 黑人巨大精品欧美一区二区蜜桃| 国产精品国产av在线观看| 国产一区二区激情短视频 | 国产亚洲最大av| 日日爽夜夜爽网站| 国产 精品1| 欧美成人午夜精品| 国产乱人偷精品视频| 亚洲视频免费观看视频| 伦理电影大哥的女人| av网站免费在线观看视频| 国产极品粉嫩免费观看在线| 国产精品欧美亚洲77777| 日日爽夜夜爽网站| 亚洲国产成人一精品久久久| 亚洲,欧美,日韩| av.在线天堂| 亚洲天堂av无毛| 中文精品一卡2卡3卡4更新| 亚洲情色 制服丝袜| 精品卡一卡二卡四卡免费| 一级黄片播放器| 男女午夜视频在线观看| 成人午夜精彩视频在线观看| 九草在线视频观看| 满18在线观看网站| 最新的欧美精品一区二区| 岛国毛片在线播放| 国产爽快片一区二区三区| 街头女战士在线观看网站| 亚洲一区中文字幕在线| 国产精品秋霞免费鲁丝片| 欧美日韩福利视频一区二区| 国产精品免费大片| 精品国产国语对白av| 超碰97精品在线观看| 赤兔流量卡办理| 黄频高清免费视频| 精品久久久久久电影网| 国产成人av激情在线播放| 国产片特级美女逼逼视频| 人人妻,人人澡人人爽秒播 | 免费黄色在线免费观看| tube8黄色片| 亚洲成色77777| 哪个播放器可以免费观看大片| 欧美另类一区| 午夜福利影视在线免费观看| 亚洲精品国产av蜜桃| 高清视频免费观看一区二区| 亚洲色图综合在线观看| 久热爱精品视频在线9| 亚洲欧美色中文字幕在线| 一级片免费观看大全| 制服诱惑二区| av网站在线播放免费| 亚洲自偷自拍图片 自拍| 亚洲av成人不卡在线观看播放网 | 精品国产乱码久久久久久小说| av天堂久久9| 97精品久久久久久久久久精品| 少妇精品久久久久久久| 国产一区二区激情短视频 | 蜜桃国产av成人99| 亚洲成人国产一区在线观看 | 国产成人精品久久二区二区91 | 日韩欧美精品免费久久| 飞空精品影院首页| 国产不卡av网站在线观看| 亚洲一码二码三码区别大吗| 亚洲av男天堂| 精品少妇内射三级| 黄色 视频免费看| 亚洲,欧美,日韩| 亚洲精品视频女| 自拍欧美九色日韩亚洲蝌蚪91| 亚洲欧美一区二区三区久久| 狠狠精品人妻久久久久久综合| 男女床上黄色一级片免费看| 日韩av在线免费看完整版不卡| 国产一区亚洲一区在线观看| 王馨瑶露胸无遮挡在线观看| 乱人伦中国视频| 女人久久www免费人成看片| 国产一区二区三区av在线| 日本猛色少妇xxxxx猛交久久| 亚洲欧美清纯卡通| 国产日韩欧美亚洲二区| 亚洲av电影在线进入| 亚洲第一区二区三区不卡| 天天躁夜夜躁狠狠久久av| 操美女的视频在线观看| 国产99久久九九免费精品| 亚洲,一卡二卡三卡| 一级a爱视频在线免费观看| 久久久久精品性色| 精品亚洲成a人片在线观看| 性高湖久久久久久久久免费观看| 观看av在线不卡| 综合色丁香网| 国产精品国产三级国产专区5o| 成年动漫av网址| 亚洲av日韩在线播放| 2018国产大陆天天弄谢| 国产精品嫩草影院av在线观看| 丝袜喷水一区| 国产av码专区亚洲av| 日日撸夜夜添| 女人爽到高潮嗷嗷叫在线视频| 免费黄网站久久成人精品| 人人妻人人澡人人爽人人夜夜| 亚洲美女视频黄频| 精品少妇内射三级| 极品少妇高潮喷水抽搐| 精品国产乱码久久久久久男人| 国产精品亚洲av一区麻豆 | 欧美日本中文国产一区发布| 69精品国产乱码久久久| av国产精品久久久久影院| 三上悠亚av全集在线观看| 伦理电影大哥的女人| 日本vs欧美在线观看视频| 天天躁日日躁夜夜躁夜夜| 精品亚洲成国产av| 你懂的网址亚洲精品在线观看| 成人国语在线视频| 免费日韩欧美在线观看| 午夜福利视频在线观看免费| 一级毛片黄色毛片免费观看视频| 七月丁香在线播放| 亚洲精品日本国产第一区| 日韩欧美一区视频在线观看| 亚洲av中文av极速乱| 国产精品女同一区二区软件| 一区二区三区激情视频| 亚洲国产日韩一区二区| 亚洲欧洲国产日韩| 中文字幕精品免费在线观看视频| 欧美在线黄色| 日韩av在线免费看完整版不卡| 久久热在线av| 久久久久久人人人人人| 久久鲁丝午夜福利片| 别揉我奶头~嗯~啊~动态视频 | 久久精品aⅴ一区二区三区四区| 制服丝袜香蕉在线| 亚洲精品乱久久久久久| 国产精品熟女久久久久浪| 在线观看免费视频网站a站| 又大又爽又粗| 免费观看性生交大片5| 久久精品国产a三级三级三级| 肉色欧美久久久久久久蜜桃| 亚洲精品国产一区二区精华液| 最近2019中文字幕mv第一页| 熟女av电影| 观看av在线不卡| av国产久精品久网站免费入址| 人人妻,人人澡人人爽秒播 | 国产一级毛片在线| 精品少妇黑人巨大在线播放| 制服诱惑二区| 一二三四中文在线观看免费高清| 亚洲国产精品国产精品| 国产免费福利视频在线观看| 夫妻午夜视频| 一级黄片播放器| 高清视频免费观看一区二区| 久久精品国产综合久久久| 久久毛片免费看一区二区三区| 菩萨蛮人人尽说江南好唐韦庄| 黄网站色视频无遮挡免费观看| 国产午夜精品一二区理论片| 最新的欧美精品一区二区| 91精品国产国语对白视频| 国产亚洲精品第一综合不卡| 男的添女的下面高潮视频| 国产成人欧美在线观看 | 99re6热这里在线精品视频| 日韩伦理黄色片| 国产在视频线精品| 精品一区二区三卡| 黄片无遮挡物在线观看| 91精品国产国语对白视频| 一级爰片在线观看| 亚洲av国产av综合av卡| 青青草视频在线视频观看| 日韩制服丝袜自拍偷拍| 制服丝袜香蕉在线| 中文字幕最新亚洲高清| 国产精品久久久久久久久免| 午夜福利免费观看在线| 中文字幕色久视频| 精品国产亚洲在线| 久久久精品欧美日韩精品| 免费不卡黄色视频| 国产精品一区二区精品视频观看| 中文字幕人妻丝袜一区二区| 欧美色欧美亚洲另类二区 | 午夜亚洲福利在线播放| 国产精品一区二区在线不卡| 国产精品影院久久| 麻豆av在线久日| 在线国产一区二区在线| 国产又爽黄色视频| 国产亚洲av嫩草精品影院| 精品久久久久久久人妻蜜臀av | 亚洲av电影在线进入| 首页视频小说图片口味搜索| 男女午夜视频在线观看| 每晚都被弄得嗷嗷叫到高潮| 久久人妻福利社区极品人妻图片| 桃色一区二区三区在线观看| 精品不卡国产一区二区三区| 90打野战视频偷拍视频| 日本三级黄在线观看| 男人舔女人的私密视频| 免费搜索国产男女视频| 黄色毛片三级朝国网站| 亚洲片人在线观看| 亚洲一卡2卡3卡4卡5卡精品中文| 天天一区二区日本电影三级 | 狂野欧美激情性xxxx| 国产午夜福利久久久久久| 日韩欧美免费精品| 人人澡人人妻人| 中文字幕人成人乱码亚洲影| 久久午夜亚洲精品久久| 在线十欧美十亚洲十日本专区| 一二三四社区在线视频社区8| 老司机午夜福利在线观看视频| 在线观看www视频免费| 99国产综合亚洲精品| 黄片播放在线免费| 巨乳人妻的诱惑在线观看| 两性夫妻黄色片| 两性午夜刺激爽爽歪歪视频在线观看 | 成人亚洲精品一区在线观看| 欧美国产日韩亚洲一区| 人人妻人人澡欧美一区二区 | 一二三四在线观看免费中文在| or卡值多少钱| av天堂久久9| 久久国产乱子伦精品免费另类| 亚洲成av人片免费观看| 不卡av一区二区三区| 欧美日本中文国产一区发布| 成熟少妇高潮喷水视频| 国产精品综合久久久久久久免费 | 国产极品粉嫩免费观看在线| av免费在线观看网站| 大陆偷拍与自拍| 精品第一国产精品| 精品国产一区二区三区四区第35| 操出白浆在线播放| 欧美精品亚洲一区二区| 黄色视频,在线免费观看| 欧美一级a爱片免费观看看 | 亚洲成a人片在线一区二区| 变态另类成人亚洲欧美熟女 | 欧美久久黑人一区二区| 亚洲国产看品久久| 亚洲人成伊人成综合网2020| 动漫黄色视频在线观看| 日韩av在线大香蕉| 美国免费a级毛片| 成人免费观看视频高清| 国产在线精品亚洲第一网站| 久久狼人影院| 首页视频小说图片口味搜索| 日韩免费av在线播放| 一区二区三区精品91| 亚洲最大成人中文| 天堂√8在线中文| 免费在线观看亚洲国产| 欧美日韩瑟瑟在线播放| 亚洲人成77777在线视频| 日日爽夜夜爽网站| 99精品欧美一区二区三区四区| 欧洲精品卡2卡3卡4卡5卡区| 18禁观看日本| 欧美人与性动交α欧美精品济南到| 国产精品98久久久久久宅男小说| 日本a在线网址| 韩国av一区二区三区四区| 免费高清视频大片| 久久伊人香网站| 国产在线观看jvid| 性少妇av在线| 亚洲专区中文字幕在线| 香蕉国产在线看| 精品福利观看| 天天添夜夜摸| 91老司机精品| 一进一出好大好爽视频| 精品午夜福利视频在线观看一区| 村上凉子中文字幕在线| 国产成人一区二区三区免费视频网站| 亚洲欧美激情在线| 午夜福利视频1000在线观看 | 成人亚洲精品一区在线观看| 欧美中文综合在线视频| 亚洲国产精品999在线| 女人精品久久久久毛片| www.自偷自拍.com| 后天国语完整版免费观看| 午夜两性在线视频| 久久精品成人免费网站| 亚洲av五月六月丁香网| 日韩欧美免费精品| 色播在线永久视频| 亚洲熟妇熟女久久| 欧美一级a爱片免费观看看 | av片东京热男人的天堂| 日韩免费av在线播放| 亚洲aⅴ乱码一区二区在线播放 | 亚洲中文av在线| 高潮久久久久久久久久久不卡| 99热只有精品国产| av天堂久久9| 久久精品国产清高在天天线| 多毛熟女@视频| 国产av一区二区精品久久| 日韩欧美在线二视频| 日韩 欧美 亚洲 中文字幕| 女人精品久久久久毛片| av中文乱码字幕在线| 很黄的视频免费| 日韩欧美国产一区二区入口| 午夜老司机福利片| 亚洲全国av大片| 中文字幕人妻熟女乱码| 久久精品91蜜桃| 国产成人影院久久av| 亚洲欧美精品综合一区二区三区| 久久久久久大精品| 人妻久久中文字幕网| 亚洲欧美激情在线| 久久久精品欧美日韩精品| bbb黄色大片| av电影中文网址| 美女免费视频网站| 淫秽高清视频在线观看| 如日韩欧美国产精品一区二区三区| 亚洲第一欧美日韩一区二区三区| 国产精品1区2区在线观看.| 国产一区二区激情短视频| 亚洲av电影不卡..在线观看| 99久久久亚洲精品蜜臀av| 国产精品久久久久久人妻精品电影| 国产1区2区3区精品| 色综合亚洲欧美另类图片| 桃色一区二区三区在线观看| 久久婷婷成人综合色麻豆| 久久久久久久久久久久大奶| 精品卡一卡二卡四卡免费| 天堂动漫精品| 国产精品永久免费网站| 无遮挡黄片免费观看| 一区在线观看完整版| 欧美精品亚洲一区二区| 免费看美女性在线毛片视频| 国产99久久九九免费精品| 黄色女人牲交| 少妇的丰满在线观看| 国产精品综合久久久久久久免费 | 色播亚洲综合网| 久久久久久久久中文| 亚洲成av片中文字幕在线观看| 欧美中文日本在线观看视频| 纯流量卡能插随身wifi吗| 国产成人影院久久av| 国产精品亚洲美女久久久|