DRIFT: Risk-Constrained Diffusion with Imitation Priors for Mixed-Autonomy Traffic Generation

2026-06-15

AI Summary

This work addresses the challenges posed by behavioral heterogeneity, tightly coupled interactions, and the generation and evaluation of high-risk rare events in mixed-autonomy traffic, where human-driven and autonomous vehicles coexist. The paper proposes the first closed-loop traffic generation framework that integrates a risk-constrained diffusion model with imitation priors. By leveraging heterogeneous perception-conditioned encoding, feasibility filtering, an online selection mechanism, and risk-aware long-tail feedback-driven adversarial alignment, the framework jointly optimizes safety, efficiency, and stability within closed-loop simulation. Experimental results demonstrate that the approach achieves superior safety-efficiency trade-offs on mixed-autonomy benchmarks and uncover the critical roles of feasibility constraints, online selection, and long-tail feedback in shaping traffic dynamics.

Abstract

Future intelligent transportation systems are envisioned to evolve toward a long-term mixed-autonomy paradigm, where human-driven vehicles (HVs) and autonomous vehicles (AVs) coexist within highly coupled traffic ecosystems. Such coexistence introduces pronounced heterogeneity, amplified uncertainty, and increasingly intricate interaction dynamics. In this context, it remains fundamentally challenging to simultaneously capture the heterogeneous behavioral distribution shifts arising from dynamic AV penetration, generate diverse yet executable trajectories under strong inter-vehicle coupling, and conduct reliable closed-loop safety and stability diagnostics for rare but high-impact events. To this end, we present risk-constrained diffusion with imitation priors (DRIFT), a mixed-autonomy traffic generation framework which unifies heterogeneity-aware conditional encoding, conditional diffusion-based executable trajectory generation, and progressive adversarial alignment enhanced by risk-aware long-tail feedback, thereby enabling traffic behaviors to be iteratively generated, filtered, selected, and validated within a closed-loop execution pipeline. In addition, a unified evaluation protocol is developed to jointly characterize safety, efficiency, and closed-loop stability across representative traffic scenarios and AV penetration regimes. Experimental results demonstrate that DRIFT achieves a strong safety-efficiency trade-off in closed-loop mixed-autonomy benchmarks, while further revealing the critical influence of candidate executability, online selection, and long-tail feedback on executable traffic evolution.

Problem

• mixed-autonomy traffic
• behavioral heterogeneity
• trajectory generation
• closed-loop safety
• distribution shift

Innovation

• conditional diffusion
• mixed-autonomy traffic
• risk-aware feedback
• executable trajectory generation
• closed-loop validation