Graph convolutional network improves freeway speed forecasting from sparse sensors
Network-Wide Freeway Speed Nowcasting and Forecasting from Sparse Fixed Sensors with a Probe-Supervised Graph Convolutional Recurrent Network
Machine LearningArtificial Intelligence
Summary
Traffic agencies often have sensors on only some roads, but they need to know speeds everywhere for now and in the near future. The authors created a new computer model that uses data from a few sensors plus historical vehicle speed data to predict current and next hour speeds on all freeway links. Their model, called GCRN+, uses a combination of smart graph-based neural networks and information about road directions to make better predictions than existing methods. This helps detect congestion more accurately even on roads without direct sensors.
What this means in practice
- •For traffic management teams: Provide accurate real-time and next-hour speed estimates on unsensored freeway links to improve traffic flow control decisions.
- •For navigation software developers: Enhance route planning algorithms by integrating more accurate near-future speed predictions for freeway links lacking direct sensor coverage.$Commercial implications: Enables transportation software companies to offer improved traffic-aware navigation services using more complete speed data.
Authors
Ran Tu, Zihao Li, Arash Mirzaei
Abstract
Traffic agencies need current and near-future speeds on every link of their network, but their fixed sensors cover only part of it. We study joint nowcasting and next-hour forecasting of hourly speeds on all 4{,}127 directional freeway links of the Dallas--Fort Worth region from sensors on 643 links (15.6%). Historical probe-vehicle data supervise every link in training; only sensor readings are used at inference. We call this setting probe-supervised sparse sensing. We propose GCRN+, a graph convolutional recurrent network that predicts deviations from a historical speed profile. It combines a parameter-free, direction-aware kernel interpolation of sensor anomalies, used as input features and as an output baseline, with a diffusion-convolutional GRU on the directed road graph, a low-rank adaptive graph and learnable link embeddings. It scales linearly with network size and trains on a CPU. Sensor readings are emulated from probe speeds, which isolates spatio-temporal inference from detector calibration. On four held-out test days, GCRN+ attains an RMSE of 2.21~mph for the current hour and 2.39~mph for the next hour on unsensored links, against 3.23~mph for the historical average. The best baselines that use the same interpolation features reach 2.53 and 2.75~mph. Adapted recent models without these features (STID, STAEformer, ImputeFormer, MoGERNN) reach at best 2.54 and 2.53~mph; this comparison is between complete pipelines rather than architectures. GCRN+ is best in every distance band except links not connected to any sensor through the road graph. Ablations within one recurrent backbone attribute most of the gain to the link embeddings and the adaptive graph, which can be trained because probe labels exist for every link. For congestion detection on unsensored links, recall rises from 77\% (historical average) and 86% (best competitor) to 89% at aprecision of 0.94.