Noise-Aware Generative Microscopic Traffic Simulation

2025-08-10

AI Summary

Microscopic traffic simulation suffers from incomplete trajectory data due to sensor perception noise, hindering accurate reproduction of real-world traffic phenomena—e.g., phantom traffic jams. Method: This paper proposes a noise-aware generative agent modeling framework. Leveraging the I-24 MOTION Scenario Dataset (I24-MSD), it explicitly incorporates sensor noise into the learning objective—rather than treating it as spurious interference—to be modeled and exploited. Integrating computer vision and generative modeling techniques, we design a noise-aware loss function enabling robust inference and optimization over incomplete trajectories. Contribution/Results: Experiments demonstrate that our model significantly outperforms conventional baselines on key traffic behavior fidelity metrics. The results validate the efficacy of the “active noise utilization” paradigm, advancing data-driven traffic simulation toward greater realism and robustness against sensing imperfections.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

Accurately modeling individual vehicle behavior in microscopic traffic simulation remains a key challenge in intelligent transportation systems, as it requires vehicles to realistically generate and respond to complex traffic phenomena such as phantom traffic jams. While traditional human driver simulation models offer computational tractability, they do so by abstracting away the very complexity that defines human driving. On the other hand, recent advances in infrastructure-mounted camera-based roadway sensing have enabled the extraction of vehicle trajectory data, presenting an opportunity to shift toward generative, agent-based models. Yet, a major bottleneck remains: most existing datasets are either overly sanitized or lack standardization, failing to reflect the noisy, imperfect nature of real-world sensing. Unlike data from vehicle-mounted sensors-which can mitigate sensing artifacts like occlusion through overlapping fields of view and sensor fusion-infrastructure-based sensors surface a messier, more practical view of challenges that traffic engineers encounter. To this end, we present the I-24 MOTION Scenario Dataset (I24-MSD)-a standardized, curated dataset designed to preserve a realistic level of sensor imperfection, embracing these errors as part of the learning problem rather than an obstacle to overcome purely from preprocessing. Drawing from noise-aware learning strategies in computer vision, we further adapt existing generative models in the autonomous driving community for I24-MSD with noise-aware loss functions. Our results show that such models not only outperform traditional baselines in realism but also benefit from explicitly engaging with, rather than suppressing, data imperfection. We view I24-MSD as a stepping stone toward a new generation of microscopic traffic simulation that embraces the real-world challenges and is better aligned with practical needs.

Problem

• Modeling realistic individual vehicle behavior in traffic simulation
• Addressing noisy and imperfect real-world traffic sensing data
• Developing noise-aware generative models for traffic simulation

Innovation

• Noise-aware generative models for traffic simulation
• Standardized dataset preserving sensor imperfections
• Adapted noise-aware loss functions from computer vision