Spatiotemporal Forecasting of Incidents and Congestion with Implications for Sustainable Traffic Control

TL;DR

This work addresses the challenge of detecting and forecasting urban traffic anomalies, such as collisions and congestion, using a reproducible SUMO-based simulation framework that generates controlled incidents. It introduces a hybrid BiLSTM–DCRNN model to localize collisions in space-time and to forecast multi-horizon edge-level metrics like travel-time index and emissions, coupling event-level detection with network-wide predictions. Key contributions include a reproducible incident-generation pipeline, aligned metric logging, and a dual-model architecture that achieves accurate collision containment and reliable spatiotemporal forecasts, demonstrated on a Broadway corridor dataset. The results underscore the potential of integrating controlled anomaly generation with deep predictive models to support anomaly-aware, sustainable traffic management in real-world networks.

Abstract

Urban traffic anomalies such as collisions and disruptions threaten the safety, efficiency, and sustainability of transportation systems. We present a simulation-based framework for modeling, detecting, and predicting such anomalies in urban networks. Using the SUMO platform, we generate reproducible rear-end and intersection crash scenarios with matched baselines, enabling controlled experimentation and comparative evaluation. We record vehicle-level travel time, speed, and emissions for edge and network-level analysis. On this dataset, we develop a hybrid forecasting architecture that combines bidirectional long short-term memory networks with a diffusion convolutional recurrent neural network to capture temporal dynamics and spatial dependencies. Our simulation studies on the Broadway corridor in New York City demonstrate the framework's ability to reproduce consistent incident conditions, quantify their effects, and provide accurate multi-horizon traffic forecasts. Our results highlight the value of combining controlled anomaly generation with deep predictive models to support reproducible evaluation and sustainable traffic management.

Figures (8)

• Figure 1: Flow chart of the modeling framework.
• Figure 2: SUMO map of NYC Broadway intersections. (a) Rear-end (AV-PV). (b) Rear-end (Bus-PV). (c) Rear-end (PV-PV). (d) Intersection (PV-PV).
• Figure 3: Stacked histogram of actual counts inside predicted $t$ range for rear-end collisions by vehicle pair.
• Figure 4: Stacked histogram of actual counts inside predicted $y$ range for rear-end collisions by vehicle pair.
• Figure 5: Stacked histogram of actual counts inside predicted $x$ range for rear-end collisions by vehicle pair.