Ultra-fast Traffic Nowcasting and Control via Differentiable Agent-based Simulation

2026-03-26

AI Summary

This work proposes the first end-to-end differentiable, large-scale agent-based traffic simulation system, overcoming the limitations of traditional fine-grained simulators that rely on inefficient gradient-free optimization due to their non-differentiability. By introducing differentiable vehicle behavior modeling, stochastic decision mechanisms, and differentiable agent interactions, the framework enables full-chain gradient-based optimization for rapid calibration and real-time control. Evaluated on the real-world Chicago road network—encompassing over one million vehicles and more than 10,000 parameters—the system achieves a 173× speedup relative to real time: calibrating against 30 minutes of observed data in 455 seconds, forecasting one hour of traffic in 21 seconds, and computing real-time control policies in 728 seconds. The entire simulation–prediction–control loop completes in under 20 minutes, substantially enhancing operational efficiency.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Sampling/Simulation-based Search
• Multiagent Systems: Agent-Based Simulation and Emergent Behavior

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Traffic digital twins, which inform policymakers of effective interventions based on large-scale, high-fidelity computational models calibrated to real-world traffic, hold promise for addressing societal challenges in our rapidly urbanizing world. However, conventional fine-grained traffic simulations are non-differentiable and typically rely on inefficient gradient-free optimization, making calibration for real-world applications computationally infeasible. Here we present a differentiable agent-based traffic simulator that enables ultra-fast model calibration, traffic nowcasting, and control on large-scale networks. We develop several differentiable computing techniques for simulating individual vehicle movements, including stochastic decision-making and inter-agent interactions, while ensuring that entire simulation trajectories remain end-to-end differentiable for efficient gradient-based optimization. On the large-scale Chicago road network, with over 10,000 calibration parameters, our model simulates more than one million vehicles at 173 times real-time speed. This ultra-fast simulation, together with efficient gradient-based optimization, enables us to complete model calibration using the previous 30 minutes of traffic data in 455 s, provide a one-hour-ahead traffic nowcast in 21 s, and solve the resulting traffic control problem in 728 s. This yields a full calibration--nowcast--control loop in under 20 minutes, leaving about 40 minutes of lead time for implementing interventions. Our work thus provides a practical computational basis for realizing traffic digital twins.

Problem

• traffic simulation
• differentiable modeling
• model calibration
• traffic nowcasting
• traffic control

Innovation

• differentiable simulation
• agent-based modeling
• traffic nowcasting
• gradient-based optimization
• traffic digital twin