Poster 20-28

Dynamic Diffusion Graph Convolutional Networks for Network Traffic Estimation

PI: Jidong J Yang

Co-PI(s): Sonny Kim, Stephan Durham, and Mi Geum Chorzepa

Institution(s): University of Georgia

Abstract

Network traffic estimation and forecasting is fundamental to transportation systems analysis and decision making in planning, design, and operation of such systems. The rapid development of intelligent transportation systems (ITS) and emerging avant-garde monitoring technologies have made traffic data collection more convenient and efficient in support of decision making at various levels. In practice, permanently located continuous count stations, complemented with portable traffic counters, are typically used for traffic data collection. However, due to budget and personnel constraints, only a limited number of segments in the road network are equipped with sensors. In other words, the network traffic flows are only partially observable, which may lead to unexpected consequences due to uncertainty and associated biases. In this study, we introduces a novel graph-based approach for estimating network-wide traffic volumes from limited sensor locations.

Please comment below with any statements or questions you may have. Also let GTI if you would be interested webinars or presentations on similar topics.