Human-Scale Quantitative Analysis on Urban Road Intersections

Road intersections have become an important part of the urban traffic system and the design quality of urban road intersections (URIs) will directly affect traffic conditions. Many evaluation methods have been established to guide the construction of urban road intersections. However, current analysis frameworks do not depict human activities in urban road space. This study proposed a sequence framework without additional investigation and experiments with several deep learning models to extract all the urban road analysis features which includes human scale variables. Three categories of URI are classified in the paper. Besides, a multivariate linear model (ML) is built to find the relationship between the possible variables and the URI distribution space. ML shows that geometric design conditions, traffic subsidiary facilities and human activities have positive effects on a good URI distribution. According to the model result, the paper proposed several suggestions to design a good URI.