Metropolis-Scale Resilient and Trustworthy Traffic Flow Inference Using Multi-Source Data

2026-05-24

AI Summary

This work addresses the challenges of city-scale traffic state inference, including sparse observations, multi-source sensing perturbations, and inter-task conflicts. The authors propose Task-Aware Attentive Neural Processes (TA-ANP), which model traffic states as stochastic processes by integrating floating car data, fixed inductive loop measurements, and road network topology. A task-aware multi-query attention mechanism is introduced to mitigate interference among subtasks, while Monte Carlo Dropout is leveraged to jointly quantify epistemic and aleatoric uncertainties. Built upon a meta-learning framework, TA-ANP rapidly adapts to changes in the sensing infrastructure. Evaluated on MMTD—the first metropolitan-scale multi-source traffic dataset—the model achieves state-of-the-art performance across all subtasks, demonstrates well-calibrated uncertainty estimates, and significantly enhances sensor deployment efficiency and system resilience under dynamic sensing network evolution.

Abstract

Inferring network-wide traffic states from sparse observations with high accuracy and trustworthy uncertainty quantification is essential for intelligent transportation systems, yet it remains challenging due to the underdetermined nature of the problem, multifaceted disturbances in sensing networks, and the inherent conflicts among multiple inference sub-tasks when modeled jointly. We propose the Task-Aware Attentive Neural Process (TA-ANP), a unified probabilistic framework for resilient and trustworthy global traffic state inference (GTSI) by fusing floating car data (FCD) with sparse fixed-detector measurements. By casting GTSI as a stochastic process, TA-ANP leverages the meta-learning properties of neural processes to adapt rapidly to changes in sensing configurations without retraining. A task-aware multi-query attention module with distinct spatiotemporal inductive biases is introduced to jointly handle three GTSI sub-tasks, while mitigating cross-task interference. For uncertainty quantification, we combine neural processes with Monte Carlo Dropout to capture both aleatoric and epistemic uncertainty. To support metropolis-scale evaluation, we construct the Metropolitan Multi-Source Traffic Dataset (MMTD), integrating fixed-loop sensor measurements, FCD statistics, and OpenStreetMap road-network data over an urban network of 2,371 road segments. Experiments on MMTD show that TA-ANP achieves state-of-the-art performance across all sub-tasks under deterministic and probabilistic metrics. The resulting well-calibrated uncertainties enable more efficient fixed-sensor placement with fewer sensor deployments. Under a Damage-Repair-Addition sensing lifecycle, TA-ANP demonstrates superior resilience in terms of disturbance absorption, performance recovery, and adaptability to unseen sensing configurations.

Problem

• traffic state inference
• uncertainty quantification
• multi-source data fusion
• sparse observations
• metropolis-scale

Innovation

• Neural Processes
• Uncertainty Quantification
• Multi-Source Traffic Fusion
• Task-Aware Attention
• Resilient Inference