Exploring the Place Dynamics of Roadside Space Conflicts with Street View Images: The Case of Beijing

Abstract

Abstract Conflicts between vehicles, cyclists, and pedestrians have been increasingly investigated as automobiles marginalize other types of mobility in crowded public spaces, particularly in roadside spaces where active mobility meets automobiles. These conflicts are extremely visible in Beijing, one of the world's most congested cities. This study explores the fined-grained spatial patterns of roadside space conflicts and factors associated with two types of conflicts (vehicle-bike, and vehicle-pedestrian). A systematic method combining spatial sampling and machine learning is described to capture the frequency of conflicts between vehicles, cyclists, and pedestrians from street view images. The ridge regression analysis identifies places with a higher probability of generating roadside space conflicts and disparate dynamics between urban and suburban/rural communities. The results of the analysis are compared with the current street regeneration plan in Beijing, which suggests that future plans should incorporate place-based roadside space management strategies in addition to the provision of on-street parking facilities.