A Generative Model for Closed-Loop Microsimulation of Signalized Intersections

2026-06-22

AI Summary

This work addresses the limitations of existing microscopic traffic simulation models in accurately capturing heterogeneous vehicle interactions at signalized intersections and the myopic, unstable behavior of learning-based trajectory predictors under closed-loop execution. To overcome these challenges, the authors propose Enactor—a vehicle-centric, generative closed-loop microscopic simulation framework. Enactor encodes dynamic agents and lane markings in polar coordinates, employs a spatiotemporally decoupled attention mechanism within a Transformer architecture to predict motion distributions, and leverages a closed-loop curriculum training strategy. The method achieves, for the first time, long-term stable generative simulation at intersections: in 4000-second closed-loop tests, it reduces the KL divergence of speed and travel time distributions by over fivefold compared to baselines, decreases red-light violations by more than an order of magnitude, and significantly outperforms constant-velocity baselines in multi-step prediction accuracy.

Abstract

Traffic microsimulators rely on hand-crafted behavior models that reproduce aggregate flow but miss the heterogeneous interactions between vehicles at signalized intersections. Learned trajectory predictors capture richer interactions but are short-horizon and tend to be unstable when run in closed loop. We present Enactor, an actor-centric generative model for closed-loop intersection microsimulation. The model focuses on vehicles; pedestrians are included as context that can influence vehicle decisions but not predicted. Dynamic actors and lane polylines are encoded in polar coordinates referenced to the intersection center. A transformer with separate spatial and temporal attention blocks predicts a distribution over each actor's next-step motion ($s$, $α$). Training uses a closed-loop curriculum so the model is exposed to its own predictions. We evaluate Enactor in two regimes. In a 4000-second simulation-in-the-loop test at two intersection geometries, Enactor controls every dynamic vehicle against a continuously refreshing actor set rather than the fixed cohort that learned trajectory predictors are usually evaluated against. It recovers the SUMO data generator's speed and travel-time distributions with KL divergence over an order of magnitude lower than a recent transformer baseline on travel time, and substantially lower on speed (roughly $5\times$ lower at Site 1), and reduces red-light violations relative to the same baseline by more than an order of magnitude. An ablation isolates the leader rear-bumper feature as the change with the largest effect on intersection-aware safety metrics. We also evaluate on real-world field data and apply the same architecture to naturalistic vehicle trajectories from a fish-eye camera at a signalized intersection and evaluate it on multi-horizon predictive tasks. Enactor outperforms a constant-velocity baseline at every horizon evaluated.

Problem

• microsimulation
• signalized intersections
• closed-loop simulation
• trajectory prediction
• heterogeneous interactions

Innovation

• closed-loop microsimulation
• generative model
• transformer with spatiotemporal attention
• polar coordinate encoding
• trajectory prediction