Toward an Integrated Cross-Urban Accident Prevention System: A Multi-Task Spatial-Temporal Learning Framework for Urban Safety Management

TL;DR

This paper tackles cross-city accident forecasting under heterogeneous, sparse, cyclical, and noisy data by proposing MLA-STNet, a unified multi-task framework that jointly learns across cities while preserving city-specific semantics. It introduces two core modules, STG-MA for spatio-temporal regularization and localization, and STS-MA for cross-city semantic modeling via adaptive graphs, fused through an adaptive channel mechanism. Empirical results on Chicago and New York City demonstrate that MLA-STNet achieves up to a $6\%$ reduction in RMSE, $8\%$ higher Recall, and $0.05$ MAP improvement, with strong robustness to input noise ($<1\%$ performance variation under moderate noise). The work offers a scalable, data-driven path toward integrated Cross-City Accident Prevention, with practical implications for urban safety management, emergency response planning, and coordinated policy making in ITS contexts.

Abstract

The development of a cross-city accident prevention system is particularly challenging due to the heterogeneity, inconsistent reporting, and inherently clustered, sparse, cyclical, and noisy nature of urban accident data. These intrinsic data properties, combined with fragmented governance and incompatible reporting standards, have long hindered the creation of an integrated, cross-city accident prevention framework. To address this gap, we propose the Mamba Local-ttention Spatial-Temporal Network MLA-STNet, a unified system that formulates accident risk prediction as a multi-task learning problem across multiple cities. MLA-STNet integrates two complementary modules: (i)the Spatio-Temporal Geographical Mamba-Attention (STG-MA), which suppresses unstable spatio-temporal fluctuations and strengthens long-range temporal dependencies; and (ii) the Spatio-Temporal Semantic Mamba-Attention (STS-MA), which mitigates cross-city heterogeneity through a shared-parameter design that jointly trains all cities while preserving individual semantic representation spaces. We validate the proposed framework through 75 experiments under two forecasting scenarios, full-day and high-frequency accident periods, using real-world datasets from New York City and Chicago. Compared with the state-of-the-art baselines, MLA-STNet achieves up to 6% lower RMSE, 8% higher Recall, and 5% higher MAP, while maintaining less than 1% performance variation under 50% input noise. These results demonstrate that MLA-STNet effectively unifies heterogeneous urban datasets within a scalable, robust, and interpretable Cross-City Accident Prevention System, paving the way for coordinated and data-driven urban safety management.

Figures (18)

• Figure 1: Overall pipeline of the Cross-City Accident Prevention System, showing the flow from cross-city datasets to the risk prediction framework and finally to prediction tasks and reasoning-related applications.
• Figure 2: Hourly accident distribution over a week in Chicago and New York City (NYC).
• Figure 3: Spatial distribution of accidents across four time periods in New York City (NYC).
• Figure 4: Spatial distribution of accidents across four time periods in Chicago.
• Figure 5: Cross-city relationship analysis between Chicago and NYC.

Theorems & Definitions (1)

• Definition 1: Multi-City Accident Risk Forecasting