Modélisation des environnements virtuels urbains dédiés à la simulation du mouvement des piétons. Modelling virtual urban environments dedicated to the simulation of pedestrian movement

Abstract

Although simulating pedestrian’s movement in urban environment is a challenging task, it has attracted the interest of researchers of different fields. In the urban planning context, movement simulation is used in the optimization of city design for ease of wayfinding and prediction of movement patterns. In wireless communication domain, predicted mobility patterns became crucial in network management optimization, allocation of critical resources and development of assisted navigation technologies. However, a realistic simulation of pedestrian wayfinding behaviour requires not only a precise modelling of human decision making process regarding strategies and heuristics applied, but also an efficient description of the spatial environment. We propose, in this thesis, a virtual urban environment model enriched by the information needed to produce realistic pedestrian movement simulation. This model describes the geometry, topology, and spatial configuration of the simulation environment. We also provide a path planning model that takes advantage of the enriched description of the environment to make virtual pedestrians show a certain behavioural realism, especially concerning their applied strategies and route choices, which are strongly affected by their familiarity with this environment. Therefore, we propose to endow each entity with a mental representation. However, our model allows creating and simultaneously managing many entities with low costs. All proposed models are inspired from spatial cognition and space syntax domains in order to achieve naturally pedestrian navigation.