LODUS A Multi-Level Framework for Simulating Environment and Population - A Contagion Experiment on a Pandemic World

2025-05-02 0 0 2.37MB 9 页 10玖币

侵权投诉

LODUS: A Multi-Level Framework for Simulating

Environment and Population - A Contagion

Experiment on a Pandemic World

Gabriel Fonseca Silva, Vin´

ıcius Cassol,

Amyr Borges Fortes Neto, Andre Antonitsch,

Diogo Schaffer, Soraia Raupp Musse

Virtual Humans Simulation Laboratory

Graduate Course in Computer Science

School of Technology

Pontiﬁcal Catholic University of Rio Grande do Sul

http://www.inf.pucrs.br/vhlab

Rodrigo de Marsillac Linn

Ofﬁce of the Chief Information Ofﬁcer

Municipal Secretary of Planning and Management

Porto Alegre City Hall - RS

https://prefeitura.poa.br/smpg

Abstract—Nowadays we are experiencing a way of life that

never existed before. The pandemic has sharply changed our

habits, customs, and behavior. In addition, a lot of work was

suddenly requested for city managers challenging them to develop

strategies to try stopping the pandemic progression. Urban

environments must be dynamic and managers need fast decisions

when working on crisis situations. In this paper we present

LODUS, a framework able to simulate urban environments on

a multi-level approach, combining macro and micro simulation

information in order to provide accurate information about

population dynamics. Furthermore, the framework LODUS is

a powerful tool when performing an urban viability study, since

the simulation results are able to highlight and predict attention

points prior to an urban environment to be built.

I. INTRODUCTION

1A smart city is an urban area that can use electronic sensors

to capture information about the population and functioning

of spaces. With this information, resources and services can

be managed to improve people’s quality of life. Besides, the

urban ecosystem is constantly evolving and it is important

to predict situations which can request immediate actions in

order to avoid issues. The use of technology is a powerful

partner when used to help designing, planing and managing a

city. In particular, simulations in virtual environments are able

to reproduce speciﬁc situations to support strategic planning

decisions.

Eid and Eldin [1] discuss that simulation systems can be

useful in the simulation of urban areas such as choice of

residential locations, problems of transport, etc. In particular,

managers are interested in the human behavior associated with

the urban planning [2]. Usually, simulation tools are speciﬁc

to contexts, for instance the mobility study and impact of

people’s life [3]. One of the challenges that urban simulation

tools present concerns with the level of detail of data and

scenarios to be simulated. For instance, if one manager wants

1Draft version made for arXiv: https://arxiv.org/

to simulate a new route in a public transportation in a city,

or inter-state, the data to be included in terms of environment

and population changes. So, it is desirable to have a framework

where managers can decide the level of detail data that should

be present in order to answer the required questions.

This paper aims to present our framework named LODUS,

to simulate, with multi level of details, a virtual urban en-

vironment taken into account the environment structure and

the population who lives on it. We propose a multi-level of

detail so the managers are able to see the information about

space with the desired detail to take decisions. In addition,

we propose and included in LODUS a model to simulate

contagion among people in the various levels of details, as

a try to help with current situation of COVID-19 pandemic

situation.

II. RELATED WORK

In this section we present the related work which supports

our research focusing on environment creation and population.

In addition, we also explore contagion related research since

the population dynamic in the world is currently being affected

by the pandemic.

Parish and M¨

uller [4] work presented the ﬁrst model for

procedurally generating different elements of a city, includ-

ing streets and parcels. The model uses an extension of L-

Systems [5] to search for local optimal successors for each

rule applied. Talton et al. [6] presented an extension of this

model to allow the parameterization of L-System rules and

terminal symbols. The model uses a Markov Chain Monte

Carlo (MCMC) method to approximate the ideal values for

each parameter to create the desired environment. However,

this work has presented limited results for larger cities, being

limited to a single point of view and requiring a well-deﬁned

likelihood function to evaluate each iteration ﬁtness.

Due to the limiting nature of L-Systems modeling of non-

organic elements, different models have been presented to

arXiv:2210.03060v1 [cs.MA] 6 Oct 2022

allow higher artistic control following the inverse procedural

modeling concept used by Talton et al. [6]. Vanegas et al. [7]

presented a model for combining behavioral modeling (e.g.

population density distribution, job distribution, high access,

land usage) and geometric modeling (e.g. road width and

number of lanes, building area, number of ﬂoors), allowing

the changes made in a certain parameter to impact on every

element of the city layout. Vanegas et al. [8] presented a

model where, given an initial city model, adjustments using an

MCMC method are made to approximate a set of parameters

(e.g. parcel average area, road curvature, building height)

deﬁned by the user. The model presented by Aliaga et al.

[9] allows the user to draw the distribution on land usage

(e.g. urban regions, agricultural regions, bodies of water) on

the base terrain to create weather simulations. Roads and

buildings for urban regions are adjusted using a Metropolis-

Hastings algorithm [10], [11], an MCMC method, to explore

the search space and ﬁnd similar results to the user-speciﬁed

weather. Speciﬁcations can include cloud coverage per region,

humidity, rain distribution, and city temperature. Mustafa et

al. [12] presented a model for the generation of urban layouts

that passively reduce water depth during ﬂood scenarios. This

work combines a procedural generation model with a hydraulic

model to evaluate water ﬂow characteristics for the created

region. A neural network is used to identify relationships

between the urban layout and the ﬂow of water during ﬂoods,

which are used as input to an MCMC model, adjusting

the environment. The evaluation takes into consideration the

desired building coverage and average water depth, etc.

Populations and crowds can be simulated with varying

levels-of-detail, with a trade-off between accuracy and compu-

tational performance. While a microscopic crowd simulation

offers individual characteristics and decision making for each

agent, it can be prohibitively computationally demanding when

simulating an environment such as an entire city. Macroscopic

crowd simulation models group up agents to reduce the

granularity of the simulation and simulate larger crowds. One

of such models is BioClouds [13], a crowd simulation model

which offers collision avoidance based on space discretization

and competition. BioClouds models crowds as clouds, i.e.,

groups of similar minded agents, which have a desire for a

certain density, speed and goal. Clouds compete for space

amongst each other, and occupy the environment in a manner

that tries to keep their desired densities respected.

If on one hand, BioClouds simulate macroscopically groups,

sometimes the microscopic simulation of people is needed.

The crisis caused by the new Corona Virus and the global

spread of COVID-19 cases turned the attention of scientiﬁc

community to the spreading disease problem. Some research

groups are focusing on SIR [14] (further formulated in Sec-

tion III-B) and SEIR [15] mathematical models to attempt

to predict when the ﬂatten of contagion curves will occur. It

is of major importance to predict those curves, given every

scenario [16]. Authorities must be able to decide when to

open schools, stores and shopping centers with the objective

to protect children education and jobs, minimizing the risk of

endangering public health.

In this paper we use the model proposed by Antonitsch [13],

where we do not simulate individuals, but groups, in a

macroscopic level. Therefore, we also propose to include in

LODUS a microscopic simulation using BioCrowds in order

to have individuals. Our goal is to customize BioCrowds

to program agents to attempt to keep recommended social

distance. Then, we extract information about social distancing

simulation to estimate contagion rates for micro-environments,

then we extrapolate those data for macroscopic simulation.

III. LODUS: LEVEL-OF-DETAIL ON URBAN SIMULATION

The ability to predict and analyze urban dynamics scenarios

is a key and distinctive support for the work of city managers

and urban planners. This paper presents LODUS, a framework

able to simulate virtual urban environments with various levels

of details. LODUS main goals is to provide information

in order to collaborate on the challenge of city planning

and management. Figure 1 illustrates the architecture of the

proposed framework.

Fig. 1. LODUS Architecture.

LODUS is a multi-level model able to be conﬁgured in two

ways: i) reproducing scenarios of real world or ii) simulating

urban dynamics before the introduction of changes in real

life. In both cases, it is important that environment and

population could be coherently simulated. The next sections

describe in details every module which compose our multi-

level framework.

A. Environment

This module allows the representation of an environment

with different levels of abstractions. At each level, a more

detailed representation of the environment may be deﬁned. As

a deeper level is included, more detailed information can be

computed. Such details store different data, such as population

distribution, road system, buildings’ geometry or even internal

buildings setups. Figure 2 presents the steps performed on the

environment construction.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

LODUS:AMulti-LevelFrameworkforSimulatingEnvironmentandPopulation-AContagionExperimentonaPandemicWorldGabrielFonsecaSilva,Vin´ciusCassol,AmyrBorgesFortesNeto,AndreAntonitsch,DiogoSchaffer,SoraiaRauppMusseVirtualHumansSimulationLaboratoryGraduateCourseinComputerScienceSchoolofTechnologyPonticalCatho...

展开>> 收起<<

LODUS A Multi-Level Framework for Simulating Environment and Population - A Contagion Experiment on a Pandemic World.pdf

共9页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

LODUS A Multi-Level Framework for Simulating Environment and Population - A Contagion Experiment on a Pandemic World

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: