DYNAMIC PEDESTRIAN ROUTE CHOICE MODEL CONSIDERING STREET LANDSCAPE AND SPATIAL CHARACTERISTICS

Abstract

In recent years, redevelopment projects and various policies have been implemented to promote compact cities where residents can live without relying on automobiles, using public transportation and walking instead. In particular, street improvements aimed at enhancing pedestrian comfort have been actively pursued. However, the quantitative evaluation of streetscape and pedestrian comfort remains a challenge. This study develops a model to describe the impact of streetscapes on pedestrian route choice behavior and evaluates street space improvements in the Toyosu district of Tokyo. Specifically, streetscape variables obtained through image analysis are incorporated into a Recursive Logit-based route choice model, and parameters are estimated from observed data. Furthermore, multiple streetscape patterns are generated using generative AI, and route choice simulations are conducted. This approach enables a quantitative evaluation that explicitly considers the relationship between streetscape and pedestrian route choice behavior, contributing to the effective planning of pedestrian-friendly urban spaces.