Match bus stops to a digital road network by the shortest path model

We investigate the bus stop matching problem that determines the associated road link of a bus stop whose latitude and longitude are available. The bus stop matching problem is challenging due to (1) the approximate representation of road maps; (2) the inaccurate location data of road maps and stops; (3) the complicated road geometry; and (4) the lack of dynamic vehicle information, such as vehicle trajectory, speed, turning, and heading. In order to exploit relational information between bus stops, we first organize the bus stops by route and list the stops in the order they appear in a given bus route. We then develop a network model that utilizes such relational information and transforms the stop matching problem into a shortest path problem. We also develop certain approaches to examine the potential mismatches by the proposed shortest path model. Case studies show that the average success rate of our approach is more than 98%. Our approach also performs robustly under a regular perturbation.