Presented at ITSC 2022
The study on “The Automatic Generation of a Road Network” by Takashi Owaki and Takashi Machida was selected for presentation at 2022 IEEE International Conference on Intelligent Transportation Systems(ITSC).
In recent years, demand for virtual road networks with characteristics of actual cities has increased in areas such as autonomous driving system development and urban design development, among others. This study proposed a method to automatically generate a road network on a city scale. The proposed method uses an iterative deep-learning model that predicts the remaining shape of roads from the partial shape of roads connected to an intersection. The road network is represented as a graph, with intersections displayed as nodes and roads displayed as links. Attributes of a city block surrounded by roads are retained as link attributes, and link attributes that constitute a city block are aggregated to determine city block attributes. In this manner, it is possible to generate a road network that includes city block attributes. We verified that using the proposed method, the entire road network of a city used for learning can be recreated with high precision and diverse road networks with partial similarities can be generated. The current result is expected to contribute to operational verification of an autonomous driving system through simulations that use generated road networks and to support road network design in urban development.
Title: Road Network Generation with City Block Attributes Using Link Attribute Aggregation
Authors: Owaki, T., Machida, T.
Appears in: 2022 IEEE International Conference on Intelligent Transportation Systems
Presented: October 5, 2022
https://doi.org/10.1109/ITSC55140.2022.9921784