We studied the ETA problem on a road network. We proposed a comprehensive and novel neural network based approach that is able to fully exploit spatio-temporal features extracted from four significant aspects: heterogeneity, proximity, periodicity and dynamicity. We built a link-connection graph to capture each route’s static contexts (i.e., spatial proximity and temporal periodicity), and we collect historical traffic conditions as its dynamic contexts.
Paper
Full text
An Effective Deep Learning Model for Route Travel Time Estimation on A Road Network
Semantic Scholar · Computer Science · 2023
Abstract
We studied the ETA problem on a road network. We proposed a comprehensive and novel neural network based approach that is able to fully exploit spatio-temporal features extracted from four significant aspects: heterogeneity, proximity, periodicity and dynamicity. We built a link-connection graph to capture each route’s static contexts (i.e., spatial proximity and temporal periodicity), and we collect historical traffic conditions as its dynamic contexts.