DRIFT: Risk-Constrained Diffusion with Imitation Priors for Mixed-Autonomy Traffic Generation
2026-06-15 • Distributed, Parallel, and Cluster Computing
Distributed, Parallel, and Cluster Computing
AI summaryⓘ
The authors created a new system called DRIFT to help simulate traffic where human-driven and self-driving cars share the road. Their system can generate realistic and safe driving behaviors by considering the mix of different vehicles and their interactions. It uses advanced methods to create and test driving paths, making sure the simulated cars behave reliably, even in rare risky situations. They also developed a way to measure how safe and efficient these simulations are across various traffic conditions. Their tests show that DRIFT balances safety and efficiency well while highlighting the importance of selecting and validating driving behaviors carefully.
mixed-autonomy trafficautonomous vehicles (AVs)human-driven vehicles (HVs)conditional diffusiontrajectory generationclosed-loop safetybehavioral heterogeneityadversarial alignmentlong-tail feedback
Authors
Yaoshen Yu, Minghui Liwang, Wenbo Zhu, Xinlei Yi, Yiguang Hong, Yuhan Su, Seyyedali Hosseinalipour
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.