CityDev, an interactive multi-agents urban model on the web

doi:10.1016/S0198-9715(02)00047-9

Computers, Environment and Urban Systems

Volume 28, Issues 1–2, January–March 2004, Pages 45-64

https://doi.org/10.1016/S0198-9715(02)00047-9 Get rights and content

Abstract

In this paper we present CityDev, an interactive multi-agents simulation model of the development of a city. The model is based on agents, goods and markets. Each agent (family, industrial firm, developer, etc.) produces goods by using other goods, and trades the goods on the markets. Each good has a price, and the monetary aspects are included in the simulation. When agents produce goods and interact in the markets, the urban fabric is built and transformed. The computer model (simulator) runs on a 3-D spatial pattern organized in cubic cells. CityDev allows interactive users to get involved in the functioning of the model. In fact, they can manage agents generated by the simulator, as well as new agents created by themselves. Agents managed by users interact with agents managed by the simulator. In addition, an administration board interactively controls the development of the city trough the urban plan, the building of new roads and the location of public facilities. In the present paper the model is described and some results are shown.

Introduction

Gaming simulation has anticipated some of the basic ideas of the multi-agent simulation approach (Hiroyuki, 2001). Usually, in urban games, the goal is to let human players learn the rules underlying the building of the urban fabric. In turn, in the multi-agent model, the main goal is the simulation of a real situation (Batty and Jiang, 1999) (Portugali, 1999). The present model, called CityDev, which was conceived for the teaching of urban planning, integrates the two approaches, namely the gaming simulation and the multi-agent, by using the web as a site where human users may interact and play with artificial agents. This particular architecture allows the model to be used both as simulation and as an interactive system. In fact, CityDev concerns both the functioning, and the planning and the management of the urban spatial-economic system. The functioning is based on the multi-agents model, while planning is enforced by an administration board. Agents can be managed both by the simulator and by human users, while the role of the administration board is performed only by human users. Hence the model may be used both as a simulation model of the urban dynamic and as a decision support system. In this latter scenario, it allows the administration board to evaluate alternative policies, and agents managed by human users, such as developers, to experiment the realization of projects concerning their core business (e.g. real estate development).

CityDev simulates the economic system of a city in its real and monetary aspects. It is based on agents, goods and markets. Each agent produces goods by using other goods and trades the produced goods in the markets. Agents are usually located in the physical space. In this simulation the evolution of the urban fabric results from the agents’ interaction. A grid of 100×100 squared cells, each 100×100 m sized, is the spatial pattern of the simulation. Built 3-D cells can be over-posed in case of a multi-floors building, up to a maximum of 10. In other words buildings of the city are considered as composed of indivisible 3-D cells (Semboloni, 2000b). The morphology of the ground and the road network are similar to that of Prato, a city of about 160 000 inhabitants near Florence (Italy), whose economy is based on textile industry.

In the following sections we first show the core structure of the model: agents’ strategies, markets, and interaction among agents. Second, we describe the role of the administration board, the phases of the simulation and the web interface. Third, we discuss the results of simulations including an example of interaction with human users. Finally, we discuss our model with respect to similar researches.

Section snippets

Agents’ strategies and goods

Agents are the subjects, the actors of the play, while goods are the objects, the basic elements which are utilized, produced and traded by agents on the markets. Agents include: families (a group of inhabitants living in the same 3-D cell), industrial firms, commercial firms, private service firms, public services, and developers. Goods include: land, labor, buildings (residential, industrial, and commercial), exported goods, imported goods, consumption goods and services.

Each agent produces a

Markets

Interaction among agents takes place in the markets. Different markets exist for different types of goods available in the simulation: land market, building market, labor market, export goods market, consumption goods market, private service market, and public services market.

On each market the sellers offer goods or services and set their prices. A good or a service is sold to the first buyer who agrees with the price. In addition, the markets are in charge of lowering the price if the good is

The economic aspect of interaction among agents

The basic parameters of the interaction among agents considered by the model concern production and consumption, salaries, and the prices of basic goods. Parameters that control production and consumption are shown in Table 2, Table 3.

In order to understand the economic interaction produced by the previous set of parameters we consider first, a developer. A developer utilizes the labor of two families, raw material and land to build a cell during a step of the simulation, which is supposed to

Administration board

The multi-agents system is managed by a group of web users playing the role of the administration board in charge of the management of the urban development (Payne & Sycara, 2000). Tasks of the board include: town planning, setting values for the strategic variables of the simulation, and the location of public services. The latter task can also be accomplished autonomously by the simulator, while the first two tasks can be performed only by the administration board.

The goal of the planning

Phases of the simulation

The model is jointly controlled by the computer simulation and by users. The simulation runs by steps. Each of these steps is supposed to represent 1 year of the real life of the city. Each step has six phases: generation and elimination of agents, management of agents by users, trading in the markets, production, update of the agents’ budget, and administration of the simulation.

During the first phase, new agents are generated by the simulator if in the previous step the demand for the output

The web interface

User interaction relies on a web interface (URL: http://fs.urba.arch.unifi.it:8080/suncity), which is an essential part of the simulation (Page, 1998, Ravid and Rafaeli, 2000). The web interface includes a home page, trough which web users can register or login. When a user enters the simulation, he or she has to choose the agents to control and consequently to perform the tasks allowed for these agents. The structure of the interface is shown in Fig. 5. In essence, a user playing the role of

Estimation of parameters

One of the positive aspect of this multi-agent model in relation to the estimation of the parameters lies in the fact that the values of some of the parameters is connected with aspects of real life. In other words one can find the price of the raw material for buildings, but it is more difficult to find in the reality the value of an exponent controlling the range of a random factor. In fact, we can identify two classes of parameters: Parameters that can be estimated independently by he model

The experiments

Experiments concern the two simulations with two different values related to evaluation of the expected profit, and a simulation in which human users have interacted with the functioning of the model. These simulations have an explorative character. The experiments consider the evolution of the city from the beginning to year 150. In fact the city grows from one seed comprising two families one industry and a commerce. The dynamic is due to a constant growth rate applied to the demand for

Relation with similar models

As explained in the introduction, the roots of this model can be found in the urban gaming simulation, and especially in the CLUG model (Feldt, 1972). Similarities with this game can be found in the economic base mechanism, in the negotiation in the market, and in the role of transportation cost in the establishment of the urban shape. The main difference lies in the fact that CityDev is totally agent-oriented, and for this reason users are obliged to play a role that has been established in

Conclusion

We have presented an interactive multi-agents model for the simulation of the urban development. Trough the interaction with human users, this model emphasizes the role of agents in the conscious orientation of the urban system. This aspect correspond to a a new necessity concerning the new role of planning in the complex self-organized system. This new role is more related to the autonomous capacity of planning of agents.

The utilization of the web enhances the possibilities of interaction

Technical aspects

CityDev is managed by a computer application based on object oriented technologies. The software is written in Java. The basic components of the CityDev interactive simulation are the simulator, the data base, the exchange module, and the Web front-end.

The simulator is the core of the simulation. It is based on a multi-agents platform and is in charge of updating the variables of the simulation, and of the management of agents not managed by human users. Geographical data are stored and

Acknowledgements

The CityDev has been funded by the University of Florence under a program for the innovation in teaching. Massimo Vassalli (National Institute of Applied Optic, Florence) collaborated at the beginning of the project with useful suggestions and ideas.

References (11)

M. Batty et al.
Multi-agent simulation: new approaches to exploring space-time dynamics within gis
CASA Working Papers
(1999)
A.G. Feldt
CLUG: community land use game. Player's manual
(1972)
Friedman, T. (1999). Semiotics of SimCity. First Monday Journal of the...
M. Hiroyuki
A gaming simulation for the social system
IPSJ SIG-Notes Intelligence and Complex Systems
(2001)
E.H. Page
The rise of web-based simulation: Implications for the high level architecture

There are more references available in the full text version of this article.

Cited by (44)

Spatial Metrics: The Static and Dynamic Perspectives
2017, Comprehensive Geographic Information Systems
Spatial metrics are quantitative measures for evaluating the physical characteristics of geographical features, such as urban settlements and structures (Reis, 2015; Reis et al., 2015). Just like other quantitative methods of spatial analysis, spatial metrics have garnered a growing interest of international research during the past two decades (Clifton et al., 2008; Liu and Yang, 2015; Schwarz, 2010). This resurgence of quantitative urban studies is attributed to several factors, such as growing concerns about sustainable development and the negative environmental consequences of urban sprawl (Huang et al., 2007; Kaza, 2013; Lord et al., 2015; Tian et al., 2014; Torrens, 2008), increasing emphasis on evidence-based policy formulation (Geertman, 2015), developments in geographical information systems and information technologies, and rising quality and availability of spatially referenced data, notably with the recent advancements in remote sensing-based methods (Clifton et al., 2008; Herold et al., 2003a; Huang et al., 2007).
Design and implementation strategy of a parallel agent-based Schelling model
2015, Computers, Environment and Urban Systems
Citation Excerpt :
Within GIScience, they have been used mainly to explore and explain dynamic changes of geospatial phenomena using simulated data. Indeed, as laboratory tools, ABMs help researchers construct theories and experiment with interactions between agents in artificial worlds (Hatna & Benenson, 2007; Ligmann-Zielinska, 2009; Semboloni, Assfalg, Armeni, Gianassi, & Marsoni, 2004). However, to simulate real-world activities with ABMs, and to verify and validate the modeling results, comprehensive data sets with detailed individual information is compulsory.
This paper illustrates the detailed design and implementation strategy of a high-performance, parallel agent-based Schelling model. There are four key issues to implement a parallel agent-based Schelling model, i.e., the ability to handle (1) the complexity of real data processing; (2) spatial data decomposition; (3) the preference and tolerance rates for multiple ethnic groups; and (4) the communication between agents. A new parallel algorithm that addresses these issues while retaining the fundamental concepts of the original sequential Schelling model was built. The parallel Schelling model was applied to the County of San Diego’s one million household agents and one million housing unit agents using land use data and 2000 Decennial Census data. Experiments using varying numbers of CPU cores demonstrate that the computational performance of the parallel Schelling model positively correlates with increased numbers of CPU cores. This paper focuses on the implementation of a parallel agent-based Schelling model. The Schelling model is parallelized to illustrate how high-performance computing can elicit better understanding of real-world geospatial phenomena. By leveraging distributed computing resources in geospatial cyberinfrastructure, the parallel Schelling model has the potential to become a crucial component in spatial decision support systems and to help policy makers understand the key trends in urban population, urban growth, and residential segregation.
A GIS-based urban simulation model for environmental health analysis
2014, Environmental Modelling and Software
Citation Excerpt :
Conceptual models have been developed since the late 19th century (von Thunen, 1826) and, over time, they have evolved into sophisticated urban simulations designed to study the evolution and design of towns and cities (Harris and Ullman, 1945), to predict their future development (Fuglsang et al., 2013) or to analyse processes occurring in urban areas (Gaube and Remesch, 2013). The most widely used models today are mathematical models that are based on the formulation of laws that control the behaviour of the system under study; examples include cellular automata (Batty, 2005; Mundia and Murayama, 2010) and multi-agent systems (Semboloni et al., 2004). An interesting alternative provide probabilistic models which assume that the laws driving the model are a set of probabilities.
This paper introduces an urban simulation model for environmental health analysis (SIENA). SIENA is a novel tool to explore urban interactions and processes with regard to exposure assessments. It is based on urban structures and relationships observed in real-world cities making it a realistic representation of a functioning city. The development of SIENA involved identifying and quantifying fundamental processes and similarities in urban areas in Great Britain and using those to guide the building of SIENA within a GIS. An internal validation confirmed SIENA's realism. Its generality, achieved through the pooling of information from different real-world cities, makes it particularly useful for developing and testing spatial epidemiological concepts and models; simulating processes and interactions relating to environmental exposure; and exploring theoretical and methodological problems in the spatial analysis of environmental health. SIENA ultimately provides a much needed tool in the form of a controlled, simplified urban simulation model.
An agent-based simulation system for evaluating gridding urban management strategies
2012, Knowledge-Based Systems
Citation Excerpt :
Besides aiming at modeling in different complex urban systems, these researches integrate some interesting methodologies and technologies. For example, an object-based approach [1] to portraying urban systems, based on the priority of spatial relationships, is proposed to deal with the problem that a regular grid of cells does not allow for direct representation of infrastructure entities and agents that are larger than a single cell; CityDev which runs on a 3-D spatial pattern organized in cubic cells is presented as an interactive ABS model of the development of a city [38]; a knowledge-based system which is composed of a global planning and coordination system is proposed for the efficient management, regulation, interactive and dynamic monitoring of urban infrastructures [35,36]; a behavioral framework is integrated into an ABS for simulating crime [29]; and an assessment framework [41] is developed, integrating spatial economic and policy decisions, energy and fuel use, and air pollution emissions and assimilation, to evaluate sustainability of the entire urban system; a integrated knowledge-based decision support system with a multilayer artificial neural network for urban planning is presented in Ref. [39]. However, no application of ABS except our early work presented in an IEEE International Conference [16], has so far been seen in the GUM field.
Traditional means of urban management are not satisfying the new demands placed by a modern city. Gridding urban management (GUM) seems to be a promising model for urban management in the 21st century. But the GUM must be more than “infrastructure” – it needs to support effective management strategies. This paper outlines our initial attempt of examining urban problems within a GUM framework and studying strategy choices. To this end, we construct an agent-based simulation system, namely DeciUrban, to explore GUM and strategy development. This paper proposes the architecture of DeciUrban and describes its core agent-based model using a model documentation protocol. A simulation of grid inspecting activities in Shanghai GUM is conducted with the purpose of assessing different inspecting strategies. The strategies involve district-first, community-first, cooperative, and random inspecting strategies. The results show how the effectiveness of strategies and the distribution of their impacts can be very different from what one would expect without the benefit of a simulation model. These simulations are presented as a demonstration that DeciUrban can help urban strategy developers understand a multitude of local interactions in cities, explore strategies, and improve community wellbeing without always designing field experiments. To sum up, simulation systems similar to DeciUrban could be valuable tools for addressing the challenges in the further development of GUM.
Modelling urban change with cellular automata: Contemporary issues and future research directions
2021, Progress in Human Geography
A framework to generate virtual cities as sandboxes for land use-transport interaction models
2021, Journal of Transport and Land Use

View all citing articles on Scopus

View full text