From a Link-Node-Based Network Representation Model to a Lane-Based Network Representation Model: Two-Dimensional Arrangements Approach

Abstract

Most mesoscopic and microscopic traffic simulation studies require the road network under analysis to be detailed at the lane level. Lane-based network representation models could serve as a more realistic platform to provide a geospatial context for traffic simulations to be performed at the level of individual vehicles in a manner that takes into consideration the geospatial context. Presently, most road networks are modeled based on a link-node structure, in which road segments are modeled as links that start, end, and connect to each other at nodes; detailed information regarding lane and lane connectivity at intersections is lost. To overcome this gap between the link-node-based and the lane-based network representations, both open source and commercial simulation packages had created specific data structures and formats, making it difficult to extend them to convert link-node representations into lane-based network representations and facilitate the sharing of resulting lane models. This paper presents an effective approach to automatically derive generic lane-based network representation models from link-node-based network representation models using the concept of two-dimensional (2D) arrangements to realize the automatic extraction of link and lane topologies. The method was prototyped and tested on a Manhattan and a Miami road network. The algorithms and tools created in this study greatly facilitate the development of mesoscopic and microscopic traffic simulation models to address a wide array of infrastructure engineering problems.