Visuospatial search in urban environment simulated by random walks

Human spatial behavior, navigation and activity in cities are affected by the visibility of urban environment. In this paper pedestrian visual search for urban locations is conceptualized as a stochastic process and modelled by random walks (a succession of random steps). Urban environments are represented as a chain of navigational decisions forming a mathematical graph of visual connectivity. Random walkers wandering on the graph imitate aggregated visual search for urban activities in the case study, which is the historical district of Tel Aviv-Yafo. The study is then compared to a set of systematically generated computational patterns represented by various visibility graph models. The main result discloses a strong positive relation between local and global visibility properties. A positive global-local relationship of spatial networks is termed intelligibility in space syntax theory and is deemed a key determinant in human navigation and wayfinding. Additionally, we compare the first passage time, an outcome of stochastic simulation, with deterministic graph theoretic measures of centrality, namely closeness and betweenness centralities. Comparison of distinct visibility types aims to identify a navigational signature of the case study and allows measurement of visual distinction between different locations in the city.