Catálogo de publicaciones - libros

When modeling land use changes in large regions or countries it has been necessary to combine two or more models operating at different scales. Typically location and relocation of population and economic activity is handled by a spatial interaction based model defined on statistical units like census tracts or counties, and the output of this model then drives a CA based model of land use, constraining cell totals in each of the regions. This approach works relatively well when the statistical areas are numerous and functionally coherent (e.g. urban centred regions). But when the areas are few, and worse, polycentric, results are very poor. An alternative approach is to attribute the activities to the cells of the corresponding land uses, and then to treat the dynamics at both scales using a single CA. In order to do this, the CA is defined with a variable size grid, so that the neighbourhood of each cell includes the entire modelled area, but the number of cells in the neighbourhood is relatively small, since the cells in each successive ring of cells in the neighbourhood is nine times as large as the cells in the preceding ring. And since each cell neighbourhood includes the entire modelled area, spatial processes at all scales are included in the cellular transition rules. The theoretical strength of the approach and the practical advantages in many applications are clear. First tests in applications previously modelled with a single scale CA linked to a spatial interaction based regional model indicate that the approach eliminates several problems inherent in the conventional approach; for example boundary effects, where urban growth cannot cross regional boundaries, disappear, and in large regions growth is distributed more realistically.

This chapter provides an overview of ongoing research projects in the DDSS research program at TUE related to multi-agents. Projects include (a) the use of multi-agent models and concepts of artificial intelligence to develop models of activity-travel behavior; (b) the use of a multi-agent model to simulate pedestrian movement in urban environments; (c) the use of multi-agent models for simulating the dynamics of land development; (d) the use of multi-agent models to simulate learning and adaptation behavior in non-stationary urban environments and under conditions of uncertainty and information search and (e) the development of computational models linking cognition, choice set formation, activity travel behavior and land use dynamics. The scope, theoretical underpinnings and results of numerical and empirical simulations are presented.

The modeling of the dynamics of settlement systems can be developed at different geographical scales according to the theoretical framework which is chosen: the micro-level of the households and entrepreneurs, the meso-level of cities and regions, the macro-level of hierarchical and spatial structures. The underlying hypotheses and the links between these three levels are discussed in the case of a multi-agent system (MAS) approach. The question of which are the driving forces of change in a settlement system is raised. Then different ways for building hybrid models combining dynamics referring to different scales are discussed. I refer to the example of SimPop, a MAS model which simulates the emergence and the evolution of a settlement system on a period of 2000 years, in order to illustrate how a function of urban governance that ensures both cognitive and decisional capacities for the evolution of cities can be introduced in a model whose rules are principally built on meso-level regularities.

New model potentials exist for coping with the complexities of today’s cities. These are related to the cognitive mediation role that modeling allows one to establish between the abstraction process (internal loop) and the external environment to which a model application belongs (external loop). The focus is turned to the two main aspects involved in that role, i.e. the modeling task and the technological interface. As far as the first is concerned, there are claims that model building in geography involves three main components: a syntactic component (how are the mechanisms underlying the functioning of the system accounted for?), a representational (semantic) component (what kind of urban descriptions are conveyed by the model?) and a purposive investigation project component (what is the modeling activity intended for?). As they increasingly rely on computing technology, models as cognitive mediators are not just simple, autonomous entities, but active complex objects. A model can therefore be understood as an ALC (Action, Learning, Communication) agent, capable of performing a certain course of Action, and permitting a certain Learning ability, which, because of its cognitive mediating role, Communicates with other kinds of agents (other models). This notion is then related to the various aspects of model-building in geography as originally introduced in the early seventies. These aspects are re-interpreted in light of the above characteristics. We conclude the paper with some remarks about the implications which may be derived as far as the harnessing of complexity in urban systems is concerned.

The ability to locate information in a complex information space requires specialized tools to support searching and browsing behavior. Inherent in browsing is the ability to navigate through informational items, while retaining a sense of orientation. A tripartite theory of navigation is presented based on cognitive studies of navigation in physical spaces, which divides navigation into three levels: planning, procedural and motor. The last two levels become critical for virtual reality, while the first two levels are critical for the traversal of more abstract information spaces. The analysis leads to various insights for information designers, which are demonstrated in two different environments. First, for hypertext navigation, it is argued that the inclusion of structural components, such as neighborhoods and landmarks, can improve the navigability of electronic spaces for browsing and non-directed search. Second, for spatial information kiosks, the use of text, images and maps, are shown to improve the accessibility of the information. Together, these two examples highlight the benefits of grounding information design in theories of wayfinding and spatial information processing.

In this paper we present an operative Wayfinding Support System (WSS) for a virtual city using the virtual model of Tel Aviv for targeted and exploration wayfinding tasks. The WSS was developed under the assumption that a design of a virtual city should allow a transfer of spatial knowledge from a real city to its virtual representation. Accordingly, the information for this system was obtained from an empirical study on Tel Aviv residents’ urban image by using their city sketch maps. The WSS uses the topological structure between the urban elements in these sketch maps to decide which elements would be highlighted to the virtual city user, according to the observed urban environment and to the user’s real time log navigation parameters (scale and perspective).

This paper aims to provide some insights into geographic environments based on our studies using various small world models. We model a geographic environment as a network of interacting objects — not only spaces, places and locations, but also vehicles and pedestrians acting on it. We demonstrate how geographic environments might be represented as a form of networks and be illustrated as small worlds. Furthermore we try to shed light on the implications of small world properties from various application perspectives.

This paper details the use of visual simulations by the Environmental Simulation Center, Ltd. (ESC) and its collaborators to involve citizens in the neighborhood/city/regional visioning and planning process. It does so by examining three visioning projects undertaken in the last three years. The case studies demonstrate how to obtain citizen input regarding their values and group identity through their participation in designing the place in which they would like to live. Fully integrating 3D/Geographic Information System-based simulations and visualizations into the visioning process makes it possible for citizens to better understand their choices at both a policy and experiential level and arrive at consensus for the future of their communities.

Planning support systems (PSS) have moved from concept to application. One of the core assumptions of PSS is that these computer-based systems can be applied in actual planning situations and found useful as decision support tools. Based on previous experiences with applied computing efforts, we need to think carefully about how best to support successful implementations of PSS. Lessons can be learned from both experiences with large-scale urban models and with the four-step urban transportation planning process. We examine the potential of PSS across four dimensions — data availability, acceptance and support, ease of use, and appropriate and useful output.