On Learning Closed-Loop Probabilistic Multi-Agent Simulator
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909715951255552 |
|---|---|
| author | Lu, Juanwu Gupta, Rohit Moradipari, Ahmadreza Han, Kyungtae Zhang, Ruqi Wang, Ziran |
| author_facet | Lu, Juanwu Gupta, Rohit Moradipari, Ahmadreza Han, Kyungtae Zhang, Ruqi Wang, Ziran |
| contents | The rapid iteration of autonomous vehicle (AV) deployments leads to increasing needs for building realistic and scalable multi-agent traffic simulators for efficient evaluation. Recent advances in this area focus on closed-loop simulators that enable generating diverse and interactive scenarios. This paper introduces Neural Interactive Agents (NIVA), a probabilistic framework for multi-agent simulation driven by a hierarchical Bayesian model that enables closed-loop, observation-conditioned simulation through autoregressive sampling from a latent, finite mixture of Gaussian distributions. We demonstrate how NIVA unifies preexisting sequence-to-sequence trajectory prediction models and emerging closed-loop simulation models trained on Next-token Prediction (NTP) from a Bayesian inference perspective. Experiments on the Waymo Open Motion Dataset demonstrate that NIVA attains competitive performance compared to the existing method while providing embellishing control over intentions and driving styles. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_00384 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | On Learning Closed-Loop Probabilistic Multi-Agent Simulator Lu, Juanwu Gupta, Rohit Moradipari, Ahmadreza Han, Kyungtae Zhang, Ruqi Wang, Ziran Robotics The rapid iteration of autonomous vehicle (AV) deployments leads to increasing needs for building realistic and scalable multi-agent traffic simulators for efficient evaluation. Recent advances in this area focus on closed-loop simulators that enable generating diverse and interactive scenarios. This paper introduces Neural Interactive Agents (NIVA), a probabilistic framework for multi-agent simulation driven by a hierarchical Bayesian model that enables closed-loop, observation-conditioned simulation through autoregressive sampling from a latent, finite mixture of Gaussian distributions. We demonstrate how NIVA unifies preexisting sequence-to-sequence trajectory prediction models and emerging closed-loop simulation models trained on Next-token Prediction (NTP) from a Bayesian inference perspective. Experiments on the Waymo Open Motion Dataset demonstrate that NIVA attains competitive performance compared to the existing method while providing embellishing control over intentions and driving styles. |
| title | On Learning Closed-Loop Probabilistic Multi-Agent Simulator |
| topic | Robotics |
| url | https://arxiv.org/abs/2508.00384 |