A method to create multiscale urban environments and associated populations

1. Introduction Augmented reality (AR) is a combination of the real scene viewed by the user and a virtual scene generated by the computer that augments the scene with additional information. There is an increasing interest for applications using AR in various domains such as military, entertainment and surgery. In all these applications the AR presented to the user enhances that person's perception of and performance in the world (Vallino 1998). On its part, Virtual Reality (VR) aims at creating totally immersive environments: the virtual world. The integration of the two adjoining approaches, AR and VR in the same application is a promising way to enhance the perception capacities of both decision makers and agents (human beings or software agents) acting in the terrain. Defence is highly interested in this connection (Piekarski 1999) and more recently, wildfire fighting was presented in (Sahli 2005) as an illustration of the cognitive complementarity between software agents and human beings in the context of a multi-agent geo-simulation. For multi-agent geo-simulations as for integrated Virtual and Augmented Realities, one of the major obstacles is the lack of models and methods to build multiscale virtual environments. Furthermore, there are various application domains which require that this multiscale virtual geographic world be populated by large populations of agents generated in a plausible way according to observations and data of similar behaviours in real settings. For specific application domains such as military and wildfire fighting the active population can be restricted to soldier troop or fire brigade who are the intervening actors. However, when we consider situations taking place in a city, the population (persons, households, firms, etc.) is large and describing behaviour becomes complex (activity, mobility, employment, services, residence, etc.); therefore we cannot focus only on specific actors like police officers.