Gordon, Richard (GM), Grimm, Donald (General Motors), Bai, Fan (Research and Development, General Motors)

Traffic Speed Prediction Using Explicit Spatial Temporal Dynamics

Scheduled for presentation during the Regular Session "Traffic prediction and estimation IV" (ThBT7), Thursday, September 26, 2024, 15:30−15:50, Salon 15

2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), September 24- 27, 2024, Edmonton, Canada

This information is tentative and subject to change. Compiled on October 14, 2024

Abstract

Modern AI and ML traffic prediction models provide accuracy, yet lack interpret-ability. Our approach designs a local neighborhood of segments and time-periods to construct a prediction data set that contain explicit dynamics across properties measured in the spatial temporal frame. The approach performs similarly to a popular TGCN ML model while providing opportunity to understand the relationship between predictor and predictand across the graph structure to which it is applied.