Search papers, labs, and topics across Lattice.
This paper introduces Flow-ERD, a multi-agent traffic simulator that balances realism and diversity in autonomous driving scenarios. By employing Agent-Type Aware Flow Matching (AFM) to ensure type-specific kinematic execution and Entropy-Regularized Distillation (ERD) to fine-tune the distribution of agent behaviors, the method effectively preserves diverse traffic patterns while maintaining realistic motion. Flow-ERD outperforms existing benchmarks, ranking first on the WOSAC test and achieving superior results on the realism-diversity Pareto front compared to reproducible baselines.
Flow-ERD achieves a groundbreaking balance of realism and diversity in traffic simulation, outperforming existing benchmarks and redefining performance metrics.
Realistic and diverse traffic simulation is essential to autonomous driving development. Yet prevailing benchmarks predominantly reward realism, and recent methods have optimized accordingly, leaving diversity underexplored. We introduce \textbf{Flow-ERD}, a multi-agent simulator that pursues realism and diversity jointly. Its backbone, \textbf{Agent-Type Aware Flow Matching} (AFM), couples flow matching's multi-modal expressiveness with type-specific kinematic execution. It preserves fine-grained diversity while keeping motions consistent with each agent type. A second stage, \textbf{Entropy-Regularized Distillation} (ERD), fine-tunes the closed-loop rollout distribution with an entropy-regularized reverse-KL objective. This mitigates covariate shift while explicitly preventing collapse onto high-density modes. We evaluate Flow-ERD with a log-free diversity metric alongside standard realism scores. Flow-ERD ranks first on the WOSAC test benchmark and dominates the realism--diversity Pareto front among reproducible baselines. Our project page is available \href{https://seulbinhwang.github.io/flow-erd-project-page/}{here}.
